growth, structural transformation, and rural change in viet nam

econstorMake Your Publications Visible.

A Service of

zbwLeibniz-InformationszentrumWirtschaftLeibniz Information Centrefor Economics

Tarp, Finn (Ed.)

Book — Published Version

Growth, Structural Transformation, and RuralChange in Viet Nam: A Rising Dragon on the Move

UNU-WIDER Studies in Development Economics

Provided in Cooperation with:Oxford University Press (OUP)

Suggested Citation: Tarp, Finn (Ed.) (2017) : Growth, Structural Transformation, and RuralChange in Viet Nam: A Rising Dragon on the Move, UNU-WIDER Studies in DevelopmentEconomics, ISBN 978-0-19-879696-1, Oxford University Press, Oxford,http://dx.doi.org/10.1093/acprof:oso/9780198796961.001.0001

This Version is available at:http://hdl.handle.net/10419/162550

Standard-Nutzungsbedingungen:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichenZwecken und zum Privatgebrauch gespeichert und kopiert werden.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielleZwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglichmachen, vertreiben oder anderweitig nutzen.

Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten,gelten abweichend von diesen Nutzungsbedingungen die in der dortgenannten Lizenz gewährten Nutzungsrechte.

Terms of use:

Documents in EconStor may be saved and copied for yourpersonal and scholarly purposes.

You are not to copy documents for public or commercialpurposes, to exhibit the documents publicly, to make thempublicly available on the internet, or to distribute or otherwiseuse the documents in public.

If the documents have been made available under an OpenContent Licence (especially Creative Commons Licences), youmay exercise further usage rights as specified in the indicatedlicence.

https://creativecommons.org/licenses/by-nc-sa/3.0/igo/

www.econstor.eu

Growth, Structural Transformation, andRural Change in Viet Nam

UNU World Institute for Development Economics Research (UNU-WIDER) was estab-lished by the United Nations University as its first research and training centre andstarted work in Helsinki, Finland, in 1985. The mandate of the institute is to undertakeapplied research and policy analysis on structural changes affecting developing andtransitional economies, to provide a forum for the advocacy of policies leading torobust, equitable, and environmentally sustainable growth, and to promote capacitystrengthening and training in the field of economic and social policy-making. Its workis carried out by staff researchers and visiting scholars in Helsinki and via networks ofcollaborating scholars and institutions around the world.

United Nations University World Institute for DevelopmentEconomics Research (UNU-WIDER)

Katajanokanlaituri 6B, 00160 Helsinki, Finlandwww.wider.unu.edu

http://www.wider.unu.edu

Growth, StructuralTransformation, and RuralChange in Viet Nam

A Rising Dragon on the Move

Edited byFinn Tarp

A study prepared by the United Nations University World Institutefor Development Economics Research (UNU-WIDER)

1

3Great Clarendon Street, Oxford, OX2 6DP,United Kingdom

Oxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,and education by publishing worldwide. Oxford is a registered trade mark ofOxford University Press in the UK and in certain other countries

The moral rights of the author have been asserted

First Edition published in 2017Impression: 1

Some rights reserved. This is an open access publication. Except where otherwise noted,this work is distributed under the terms of a Creative Commons Attribution-NonCommercial-Share Alike 3.0 IGO licence (CC BY-NC-SA 3.0 IGO), a copy of whichis available at https://creativecommons.org/licenses/by-nc-sa/3.0/igo/.

It is permitted to reuse, share and adapt this work, subject to the following terms:

Attribution - appropriate credit is given to the original work, the copyright holder andcreator, and any changes made to the work are properly indicated.

Non-Commercial - thework, or any adaptation of thework,may not be used, distributedor reproduced in any format, by any means, for commercial purposes

Share-Alike - the work, or any adaptation of the work is distributed under the samelicence terms as the original , with a URL link provided to the licence.

Enquiries concerning use outside the terms of the Creative Commons licence shouldbe sent to the Rights Department, Oxford University Press, at the above address or [email protected].

Published in the United States of America by Oxford University Press198 Madison Avenue, New York, NY 10016, United States of America

British Library Cataloguing in Publication DataData available

Library of Congress Control Number: 2016945575

ISBN 978–0–19–879696–1

Printed in Great Britain byClays Ltd, St Ives plc

Links to third party websites are provided by Oxford in good faith andfor information only. Oxford disclaims any responsibility for the materialscontained in any third party website referenced in this work.

OUP UNCORRECTED PROOF – FIRST PROOF, 9/2/2017, SPi

© United Nations University World Institute for Development Economics Research(UNU-WIDER) 2017

https://creativecommons.org/licenses/by-nc-sa/3.0/igo/

Foreword

I arrived in Viet Nam for the first time in August of 2000 to start up a Danida-funded programme of development research and capacity-building at theCentral Institute of Economic Management (CIEM) of the Ministry of Plan-ning and Investment (MPI) in Hanoi. At the time I was amid-career Universityof Copenhagen associate professor on the brink of entering my fifties. Littledid I know that the engagement with CIEM and Viet Namwould lead to morethan fifteen years of intense collaboration. They beganwith almost three yearsof residence in Hanoi, which were followed by some fifty study visits, eachranging from one to several weeks over a period of twelve years. My profes-sional field experience in development economics was until 2000mainly fromsub-Saharan Africa, so I was eager to engage and get to know my new Asian‘home’—seen by many as an emerging tiger. Soon after my arrival I stoppedreferring to Viet Nam as a tiger.

A well-known Vietnamese colleague, Dr Vo Tri Thanh, laughed whenI asked about his view. He added that maybe Viet Nam is a tiger—but at besta tiger that is making the transition from using a bicycle to riding a motorbike!This picture has been sticking in my mind ever since, and I gradually came tothink of Viet Nam as a rising dragon. A dragon that somehow moves differ-ently from a tiger. Eager, yet more careful, as another close CIEM colleague(Ms Vu Xuan Nguyet Hong) has argued. It also became clear early on, as statedin our very first project report, that:

The process of economic reform in Viet Nam can be compared to travelling a long,winding road through dangerousmountains and huge river valleys. Great achieve-ments have beenmade since Doi Moi was initiated in 1986, but Viet Nam has onlycome part of the way to overcoming the dual challenges of poverty and under-development. Major challenges lie ahead . . .

This was manifestly the case in relation to the generation, availability, and useof good-quality data. Without quality data it is impossible to produce academ-ically sound, yet practical and relevant evidence-based policy advice in anincreasingly global and competitive economic environment. Helping fill thisgap has, over the years, been the number one priority in the CIEM–Danidacollaborative programme. We were therefore proud to publish the first

Vietnamese Social Accounting Matrix (SAM) in 2001 in support of economy-wide policy design and implementation. It provided a much needed macro-economic map, which has since been updated frequently. Such a map is—asI well knew from my African experiences—an indispensable tool in anymodern economy-wide analysis trying to take account of supply-and-demandbehaviour and the role of market institutions.The SAMwork was highly effective in other important ways. It helped bring

into focus an even bigger gap in the available data in Viet Nam, namely thecrucial need to come to grips with themicroeconomic situation and behaviourof households and enterprises, including their access to and interaction withkey markets, especially in the poorer rural areas. To illustrate, this gap can becompared to generating the critically important specifics of a bigger macro-economicmapwithout which studies of growth and structural transformationhave little concrete to say about the lives of real people.Many developing countries—Viet Nam included—continue to struggle to

raise incomes per capita, and a large number of them have, over the past fewdecades, succeeded in generating significant (albeit not always stable) growth.A common feature of the convergence of these low-income countries is afundamental change in the pattern of economic activity, as households reallo-cate labour from traditional agriculture to more productive forms of agricul-ture and modern industrial and service sectors. The combination of theselarge-scale shifts in work and labour allocation and the resulting changes inthe composition of economic output are collectively referred to as the struc-tural transformation of the economy. A better understanding of what thisprocess means for the welfare and socioeconomic characteristics of the ruralpoor is essential. This is the case for both the development profession andpolicy makers at large in coming to grips with the task of promoting equitableand sustainable development and ending poverty. I note that this aspiration isalso the key objective in the 2030 Sustainable Development Goals (SDGs),adopted by the international community at the UN General Assembly inSeptember of 2015—but here I am getting ahead of myself.The origin of this volume is muchmore down to earth. It dates back to 2002

when the first pilot Viet Nam Access to Resources Household Survey (VARHS),covering some 930households, was carried out. The results of VARHS02 in turninspiredCIEMand theCentre for Agricultural PolicyConsulting of the Instituteof Policy and Strategy for Agriculture and Rural Development (CAP-IPSARD) ofthe Ministry of Agriculture and Rural Development (MARD), the Institute ofLabour Science and Social Affairs (ILSSA) of the Ministry of Labour, Invalidsand Social Affairs (MOLISA), and the Development Economics ResearchGroup (DERG) of the University of Copenhagen, together with Danida, toplan and carry out a more ambitious VARHS in 2006 to increase coverage andprovincial representativeness. Since then, the survey of these households

vi

Foreword

has been carried out every two years, that is, in 2008, 2010, 2012, and 2014. It ison this basis the present volume builds, and the 2016 survey is getting ready tomove into the field under the auspices of UNU-WIDER as I complete thisforeword.

Importantly, since the VARHS has surveyed the same rural households overtime, it is by now a very strong tool for gaining detailed and policy-relevantinformation about the economy and society of rural Viet Nam. In economicterminology, the VARHS includes a truly unique 2006–14 balanced panelsurvey of the changing life and work of rural families across the country.While five detailed descriptive cross-section reports for each of the surveyyears are available, this volume presents, for the first time, a comprehensiveset of detailed analytical studies where we rely throughout on the coherentdata from the 2,162 households from 466 communes that (as furtherdescribed in Chapter 2) make up the balanced 2006–14 VARHS panel; atten-tion is focused here on the time dimension rather than individual cross-section information. In other words, all chapters—except for the frameworksetting introduction in Chapter 1 and, to some extent, Chapter 12—relyextensively on this VARHS panel; the individuals in the households includedin this panel have all lived through and experienced a critical period in VietNam’s economic development process while managing their personal andhousehold lives. How they coped and ended up performing in a highlydynamic macroeconomic environment is key in what we try to uncover.

The fieldwork behind the series of VARHS consisted of detailed anddemanding interviews carried out, under often stressful conditions, in themonths of June and July in each round in the rural areas of twelve provincesin Viet Nam as follows: (i) four (ex-Ha Tay, Nghe An, Khanh Hoa, and LamDong) were supported by Danida under its Business Sector ProgrammeSupport (BSPS); (ii) five (Dak Lak, Dak, Nong, Lao Cai, Dien Bien, and LaiChau) received assistance under the Agriculture and Rural DevelopmentSector Programme Support (ARDSPS); and (iii) three (Phu Tho, Quang Nam,and Long An) were all initially surveyed in 2002 andmore recently covered bythe BSPS. The location of these twelve provinces are shown on the mapsprovided in Chapter 2.

ILSSA carried out the wide range of tasks related to the planning andimplementation of the VARHS in the field, while DERG and, later on, UNU-WIDER collaborated with CIEM and IPSARD in all aspects of survey design anddata analysis. A full package of capacity-building activities by DERG and UNU-WIDER staff, including formal courses, on-the-job training, and a wealth ofseminars, were conducted in Viet Nam, in Denmark, and elsewhere through-out this process, under evolving institutional collaborative arrangements. Theshared aim was to ensure that the VARHS project developed both the datarequired to deliver policy-relevant research to decision makers and the

vii

Foreword

research capacity within Vietnamese institutions to take advantage of thatdata.I wish to highlight in particular that VARHS was designed from the very

beginning as a collaborative research effort. Another explicit objective was tocomplement the nationally representative Viet Nam Household Living Stand-ards Survey (VHLSS) conducted biennially by the General Statistics Office(GSO). Many households surveyed in the VARHS have also been surveyed inthe VHLSS. Importantly, rather than focusing on estimating consumptionpoverty rates, a key objective of the VHLSS, the VARHS has, throughout,been targeted at gathering high-quality data about issues such as saving,investment, land use, interaction with formal and informal markets, andparticipation in rural institutions and rural social structure. More specifically,the VARHS includes an extensive number of ethnic and rural poor householdsthat have been relatively excluded from traditional growth processes. Thismeans that the evidence from VARHS can support the identification of pol-icies for inclusive growth that leaves no group or minority behind, closely inline with international calls for a data revolution within the context of the2030 sustainable development agenda referred to earlier in this foreword.To be sure, I did not foresee in 2000 that the report of the UN Secretary

General’s High-Level Panel of Eminent Persons (HLP) on the Post-2015 Devel-opment Agenda entitled A New Global Partnership: Eradicate Poverty and Trans-form Economies through Sustainable Development, would call, some fifteen yearslater, for a data revolution for sustainable development post-2015 as follows:

We also call for a data revolution for sustainable development, with a new inter-national initiative to improve the quality of statistics and information available tocitizens. We should actively take advantage of new technology, crowd sourcing,and improved connectivity to empower people with information on the progresstowards the targets.

As Director of the United Nations UniversityWorld Institute for DevelopmentEconomics Research (UNU-WIDER) since 2009 and, in this capacity, in recentyears as a member of the UN Task Team for the formulation of the post-2015development agenda, I have come to appreciate these demands for inter-national action. The HLP call for a data revolution is most pertinent, andI note that while substantial improvements in statistical systems have beenregistered in many developing countries over the past two decades, muchremains to be done in many sectors and countries. The HLP notes that morethan forty countries lack sufficiently strong systems to properly track trends inpoverty; and the panel also notes unsatisfactorily high time lags for reportingMDG (Millennium Development Goals) outcomes.Recently, large-scale revisions of gross domestic product (GDP) estimates in

Ghana and Nigeria as well as elsewhere serve as reminders of broad-based

viii

Foreword

weaknesses in statistical systems that persist across the developing world,including both Africa and the Asia-Pacific region. With regard to thisbackground—and recalling UNU-WIDER’s long-standing expertise in innov-ation in data collection and analysis—I am led to strongly confirm the viewthat data will be at the centre of development action in the coming years.

At the same time, while the logic of a concerted push towards a ‘datarevolution’ is compelling, these calls are often rather vague—and it is indeednot entirely clear from ongoing debates that it is widely understood what sucha revolution actually requires and means in concrete practice.

The aims of this volume were formulated with these concerns in mind,using Viet Nam as a case study, due to the concrete and unique, yet somewhatcoincidental, availability of the solid VARHS experience and panel data set.Furthermore, Viet Nam’s contemporary similarities to a large number ofdeveloping economies make its experience and policy recommendations,based on analysis of microeconomic data, highly relevant for many regionaland extra-regional stakeholders. In fact, Viet Nam provides an exceptionallyinformative environment in which to observe and consider the economic andsocial mechanisms underlying:

• a rural economy in transformation,• the critical importance of key production factors and institutions, and• the complex set of welfare outcomes and distributional issues.

These dimensions therefore make up the three component parts of thisvolume, identifying throughout the associated policy challenges after settingthe scene in the introductory Chapters 1 and 2, and laying out a series ofpolicy implications in the concluding Chapter 14. In my assessment theinsights from this experience should be taken to heart and considered care-fully in other countries and development partnerships when developing 2030SDG strategies and actions in search of inclusive development and the aspir-ational goal of leaving no one behind.

In sum, the aims of this volume are to:

• Provide an in-depth evaluation of the development of rural life in VietNam over the past decade, combining a unique primary source of paneldata with the best analytical tools available.

• Generate a comprehensive understanding of the impact of rural house-hold access to markets for land, labour, and capital, on the one hand, andgovernment policies on growth, inequality, and poverty at the villagelevel in Viet Nam, on the other, including the distribution of gains andlosses from economic growth.

• Serve as a lens through which other countries and the internationaldevelopment community at large may wish to approach the massive

ix

Foreword

task of pursuing a meaningful data revolution as an integral element ofthe 2030 sustainable development agenda.

• Make available a comprehensive set of materials and studies of use toacademics, students, and development practitioners interested in an inte-grated approach to the study of growth, structural transformation, andthe microeconomic analysis of development in a fascinating developingcountry.

I hope with this volume to provide a comprehensive analytic contributionto a crucial topic within the discipline of development economics based onfifteen years of continued in-country efforts. I also hope this volume can helppersuade national and international policy makers (including donors) of theneed to take the call for a data revolution seriously, in rhetoric, in concreteplans and budget allocations, and in the necessary sustained action at countrylevel. This is where inclusive socioeconomic development is needed to benefitpoor and discriminated people, who are struggling to make ends meet.

Finn TarpHelsinki, October 2016

x

Foreword

Acknowledgements

The intention of putting together this volume developed gradually over aperiod of more than a decade. A significant number of people have workedtogether with me in many capacities during the planning, implementation,and analysis of the VARHS. While I will, in what follows, try to do justice totheir many vital contributions, I apologise upfront for any omissions. The listis long and a complete inventory is simply not feasible for reasons of space.

A profound debt is owed to senior colleagues in Viet Nam, including theformer Presidents of CIEM, Dr Le Dang Doanh, Dr Dinh Van An, and AssociateProfessor Le Xuan Ba, as well as the current CIEM President Dr Nguyen DinhCung. Together with the former Director General of IPSARD, Dr Dang KimSon, the present Director, Dr Nguyen Do Anh Tuan, the two former and thepresent Director of ILSSA, respectively Dr Nguyen Huu Dzung, Dr Nguyen ThiLan Huong, and Dr Dao Quang Vinh, they worked directly withme in guidingthe VARHS effort from beginning to end of the five survey rounds. I have, inthis way, come to appreciate the key leadership qualities that have helpedpromote effective collaboration between all partners in VARHS. These top-level colleagues also contributed in critical ways to the very many seminars,workshops, and conferences that have been an integral part of the VARHSprocess, and which are fully documented on CIEM’s website.

Financial support from Danida under its various programmes over theperiod in reference is recognized with sincere gratitude. A particular thanksis due to the former Danish ambassador in Viet Nam, H.E. Peter Lysholt-Hansen. Peter was—with his never-failing sense of strategic priorities—highlyinstrumental in the early stages of setting up the VARHS, and without thissupport the VARHS would never have seen the light of day. Ambassador JohnNielsen followed and supported the research effort until the end of Danidasupport in 2014.

I hasten to add that our work would not have been possible without con-tinuous professional and administrative interaction, advice, and encourage-ment from a large number of individuals at CIEM and IPSARD. Among manyothers, I would like to highlight my gratitude to the following.

From CIEM, the former Vice-President, Mrs Vu Xuan Nguyet Hong, andpresent Vice-President, Dr Nguyen Tue Anh, have been close collaborators

from the very beginning, and former Director of the Agriculture and RuralDevelopment Policy Research Department, Dr Chu Tien Quang, and Dr DangThu Hoai, who is now one of CIEM’s directors, provided key inputs in theearly stages of our work. The same goes for the former CIEM research partnerteam including Mr Luu Duc Khai, the present Director of the Agriculture andRural Development Department, and Mr Nguyen Huu Tho, Ms Hoang XuanDiem, and Mrs Le Thi Xuan Quynh. Moreover, I am most grateful to ProjectAssistants Ms Do Hong Giang and Ms Bui Phuong Lien. Without their tirelesssupport in the organization of numerous project activities, including thepublication of countless reports and studies during a decade of project work,the present volume would have been impossible.From IPSARD, a particular thanks goes to Dr NguyenNgoc Que,Mrs Nguyen

Le Hoa, Ms Nguyen Thi Ngoc Linh, Mrs Pham Thi Phuong Lien, Ms Do LienHuong, Ms Tran Thi Thanh Nhan, Mr Ngo Quang Thanh, Ms Hien Pham, andMr Do Huy Thiep. They all contributed to our many studies along the way,and were supported by Mrs Tran Thi Quynh Chi and Mr Phung Duc Tung onthe programming and administrative side.Turning to the highly productive and stimulating collaborationwith the data

collection and management teams from ILSSA, they were effectively coordin-ated by the former Directors, Dr Nguyen Huu Dzung and Dr Nguyen Thi LanHuong; the present Director, Dr Dao Quang Vinh; Vice Director, Mr Le NguBinh; and their immediate colleagues including Ms Chu Thi Lan, Ms NguyenHai Ninh, Ms Nguyen Phuong Tra Mi, Mr Luu Quang Tuan, Ms Hoang ThiMinh, Ms Le Quynh Huong, Mr Le Hoang Dzung, Mr Nguyen Tien Quyet, MrNguyenVanDu, andMs Tran ThuHang. ILLSA alsomanaged the coordinationwith the GSO and Dr Nguyen Phong, who provided early, most appreciated,advice on sampling issues. The survey would not have taken off without theefforts of these and many other ILSSA staff, too numerous to name here, incompiling the questionnaires, training enumerators, implementing the surveyin the field, and cleaning the data. A particular thanks is also extended for theeffective and most helpful assistance provided by all administrative levels inVietNamfrom the centre inHanoi andall thewayout to the provincial, district,and commune officials and people who helped organizing numerous field tripsand pilots over more than a decade of work. Without these crucial effortsneither I personally nor the international collaborators in general would havebeen in a position to even begin comprehending the realities and challenges ofrural life in Viet Nam. I hope all we learnt comes out clearly.Importantly, I would like to express my deepest sense of appreciation for

the valuable time the several thousand rural households in twelve provincesof Viet Nam made available to us in 2006, 2008, 2010, 2012, and 2014 duringthe interviews carried out as part of this study. It was a humbling and thought-provoking experience to see the openness and eagerness with which they

xii

Acknowledgements

welcomed and engaged with us all and the many enumerator teams, andI sincerely hope that the present volume will prove useful in the shared searchfor effective policies geared towards improving their daily livelihoods. This is,in the final analysis, the overarching goal of this work, and my own personalambition.

To the many staff at the Danish Embassy, who have supported us underthe guidance of the ambassadors mentioned, I would like to acknowledge theefforts of former Deputy Heads of Mission, Tove Degnbol and Lis Rosenholm,alongside Mimi Grønbech, Henrik Vistisen, Cathrine Dolleris, and AndersBaltzer Jørgensen, as well as Danida advisor Ole Sparre Pedersen. A veryparticular set of thanks goes to Ms Vu Huong Mai, who, together withMs Nguyen Thi Thu Hang, Mr Hoang Van Tu, Mrs Nguyen Thi Phuong Bac,and Ms Nguyen Thi Phuong Thao, provided much of the essential adminis-trative support and oversight required from the Danish Embassy.

Each VARHS round involved careful preparation, implementation, analysis,and presentation and discussion of results in a wide range of workshops andlaunching events with a large number of participants. The present volumebenefited from the specific insights and helpful comments made by Dr LeDang Doanh, Dr Vu Thi Minh, and Ms Nguyen Thi Kim Dzung at a nationalworkshop held in Hanoi at CIEM on 19 May 2015 where the first draft of thisvolume was presented; and the media attention the VARHS has attracted overthe years in Viet Nam is also acknowledged.

Furthermore, I would like to mention some additional international col-leagues, who, at different stages of the VARHS process, provided most helpfuladvice and support. They include Carl Kalapesi, AdamMcCarty, and the staff atMekong Economics Ltd, who worked with me on VARHS02; Phil Abbott whomade available his wealth of insights into rural development and questionnairedesign; and SarahBales andBobBaulch,who shared theirmost valuable insightsfrom Viet Nam at critical stages of the work. Mikkel Barslund and Katleeen Vanden Broeck also worked with us in the early stages of the VARHS (in Viet Namand in Copenhagen) as did Lotte Isager; while Simon McCoy and Theo Talbotprovided essential programme, practical, and many other kinds of supportduring their respective stays in Hanoi based at CIEM in ‘my old office’. Thanksare due as well to the administrative and secretarial staffs at the University ofCopenhagen, too numerous to be listed here, although I do feel a need to thankChristel Brink Hansen in particular for her efforts over so many years. I feelconfident theyare all aware ofhowmuchandhowdeeply I appreciate their dailyefforts in making the VARHS happen.

I now turn to the individual authors, who have contributed so effectivelyto this edited volume. Their profiles can be found in the list of contributors,and I wish to state my admiration and gratitude for their analytical work andquantitative research that makes up the very core of this volume. Particular

xiii

Acknowledgements

thanks in this group go to seven team-mates. They have been close partners inthe VARHS effort and in so much other work: Irish Trinity College professor,Carol Newman, and Danish University of Copenhagen professor, ThomasMarkussen, with whom I have had marvellous working relationships formore than a decade, two other outstanding University of Copenhagen asso-ciates, Ulrik Beck and Kasper Brandt, a very promising UNU-WIDER researchcolleague Saurabh Singhal, and a contributor to both this volume and otherwork, University of Sussex professor, AndyMcKay. This group of people was—together with the other contributors—instrumental not only in workingrelentlessly with me in putting the VARHS data together and making senseof it all. They also collaborated in the very many concrete practical andanalytical challenges involved in the VARHS process necessary to producethis volume. The productivity of our professional partnership has been provenover and over again by the truly respectable research output that has appearedin some of the best development journals around. I am committed to doingmy utmost to ensure that the mutual collaboration we have managed toestablish will continue to develop and flourish.I hope it is clear that I am indebted to a large number of people who offered

critique and most helpful comments. They include as well, and very import-antly, Oxford University Press’s Economics Commissioning Editor AdamSwallow and his team, and four anonymous referees. Your many encourage-ments, professional guidance, and constructive critique were essential inhelping to sharpen the research questions and the approach adopted in thisvolume. I wish to convey my most sincere gratitude.UNU-WIDER and its dedicated staff provided steady support in producing

this volume, including excellent research assistance by Risto Rönkkö andSinnikka Parviainen on Chapter 1. Particular thanks go to Lorraine Telfer-Taivainen for all of her careful, critically needed, and sustained editorial andpublication support, including the many contacts with OUP; and Anna-MariVesterinen and the group of copy editors (first of all Lesley Ellen) for helping inputting out the many UNU-WIDER working papers produced during thecourse of the research underpinning this volume.UNU-WIDER’s work is supported by the governments of Denmark, Finland,

Sweden, and theUnitedKingdomas coredonors, and in this case a special projectcontribution was additionally provided by the Korea International CooperationAgency (KOICA). UNU-WIDER gratefully acknowledges this vital research fund-ing without which this book would not have seen the light of day.Finally, and importantly, while advice has been received from all these and

many more colleagues and friends, I take full responsibility for any remainingerrors or shortcomings in interpretation. All the usual caveats apply.

Finn TarpHelsinki, October 2016

xiv

Acknowledgements

Contents

List of Figures xviiList of Tables xxiList of Abbreviations xxvNotes on Contributors xxvii

1. Viet Nam: Setting the Scene 1Finn Tarp

2. Characteristics of the VARHS Data and Other Data Sources 26Kasper Brandt and Finn Tarp

Part I. A Rural Economy Transformation

3. Local Transformation: A Commune-Level Analysis 51Ulrik Beck

4. Commercialization in Agriculture, 2006–14 68Chiara Cazzuffi, Andy McKay, and Emilie Perge

5. The Rural Non-Farm Economy 91Christina Kinghan and Carol Newman

Part II. Key Production Factors and Institutions

6. Land Issues: Markets, Property Rights, and Investment 117Thomas Markussen

7. Labour and Migration 139Gaia Narciso

8. Information and Communication Technology 158Heidi Kaila

9. Social and Political Capital 181Thomas Markussen

Part III. Welfare Outcomes and Distribution Issues

10. Welfare Dynamics: 2006–14 205Andy McKay and Finn Tarp

11. Gender Inequality and the Empowerment of Women 222Carol Newman

12. Children and Youths 237Gaia Narciso and Carol Newman

13. Ethnic Disadvantage: Evidence Using Panel Data 256Saurabh Singhal and Ulrik Beck

Part IV. Lessons and Policy

14. Lessons Learnt and Policy Implications 279Finn Tarp

Index 295

xvi

Contents

List of Figures

1.1 Real GDP growth, Viet Nam 3

1.2 Real GDP per capita growth in selected countries, 1985–2013 4

1.3 Real GDP per capita in Southeast Asian countries 4

1.4 Sectorial distribution of aggregate Vietnamese output 5

1.5 Agriculture value added per worker (constant 2005 US$) 6

1.6 Fixed (wired) broadband subscriptions (per 100 people), 2006–13 6

1.7 Share of population aged 15–64 years (% of total population) 7

1.8 Employment of 15-year-olds and older to total population (in %) 8

1.9 Inflation in Viet Nam (annual changes in % in the CPI) 8

1.10 Inflation in selected countries (annual changes in % in the CPI) 9

1.11 Monetary policy interest rate (Viet Nam) 9

1.12 Domestic credit provided by financial sector (% of GDP) 11

1.13 Domestic credit to private sector (% of GDP) 11

1.14 Trade (exports plus imports) as a share of GDP (%) 12

1.15 Trade balance as a share of GDP (%) 13

1.16 Current account balance as a share of GDP (%) 13

1.17 Foreign direct investment, net inflows (% of GDP) 14

1.18 Total reserves excluding gold as a share of GDP (%) 14

1.19 US dollar/Vietnamese Dong exchange rate 15

1.20 Household final consumption expenditure (% annual growth) 16

1.21 Life expectancy at birth, female (years), 2006–13 16

1.22 Life expectancy at birth, male (years), 2006–13 17

1.23 Prevalence of undernourishment (% of population) 18

1.24 Depth of the food deficit (kilocalories per person per day) 18

1.25 Poverty headcount ratio at US$1.25 a day (PPP) (% of population) 20

2.1 Map of Viet Nam 28

2.2 Location of the twelve VARHS provinces 29

2.3 The VARHS communes 30

2.4 Share of households with a male as household head in 2014 34

2.5 Average age of household head in 2014 35

2.6 Percentage of Kinh households in balanced sample in 2014 36

2.7 Average years of schooling for all household members aged15 or above in 2014 37

2.8 Average years of schooling for household heads in 2014 38

2.9 Rotating sampling design of VHLSS 43

3.1 Average number of households in VARHS communes by region 53

3.2 Most important occupations by year, % of communes 54

3.3 Most important occupations by region in 2014, % of communes 55

3.4 Average land use shares by year and region 56

3.5 Presence of six commune facilities over time, % of communes 59

3.6 Distances to transportation and other facilities by year, % of communes 60

3.7 Distances to transportation and other facilities in 2014 by region,% of communes 61

3.8 Percentage of communes with streetlights and drinking waterdistribution network by region and over time 61

3.9 Percentage of communes with at least one internet access pointby region and over time 62

3.10 Percentage of communes affected by different commune problemsin the past, present, and future 64

3.11 Percentage of communes affected by different problemsin 2014 by region 65

4.1 Rice production in Viet Nam, 1975–2014 (tonnes) 69

4.2 Some summary characteristics relating to commercializationfor the full sample 73

6.1 Landlessness 119

6.2 Farm size 121

6.3 Number of plots operated, by region 122

6.4 Land purchases and sales in the last two years, by region 124

6.5 Land purchases and sales in the last two years, by income quintile 125

6.6 Share of households renting land in and out, by region 126

6.7 Share of households renting land in and out, by income quintile 127

6.8 Land-use certificates 129

8.1 Geographical distributions of technology ownership 161

8.2 Number of phones owned by household 163

8.3 Sources of internet access in 2006 and 2014 168

xviii

List of Figures

8.4 Most important sources of information, 2014 176

9.1 Communist Party membership 183

9.2 Mass organization membership 185

9.3 Membership of mass organizations and other voluntary groups 186

9.4 Non-mass organization membership 187

9.5 Generalized trust and mistrust 189

9.6 Generalized trust 190

9.7 Share of financial helpers who are relatives of the respondent 191

9.8 Share of rented plots where the tenant is a relative of the landlord 192

10.1 Kernel density plots of different welfare measures 208

13.1 Evolution of monthly food expenditure (a) and income (b) by ethnicityin real 1000 VND, 2006–14 259

13.2 Non-parametric estimates of Kinh and non-Kinh income growth,depending on initial income, 2008 and 2014 260

13.3 Household asset ownership rates by asset and ethnicity, 2006–14 261

13.4 Income diversification, 2006–14 264

13.5 Land quality and redbook ownership, 2014 265

13.6 Most important constraints before harvest as reported by householdsby year, 2008–14 266

13.7 Most important constraints after harvest as reported by householdsby year, 2008–14 267

13.8 Access to credit by ethnicity, 2008–14 268

13.9 Additional distances for non-Kinh households by year, 2008–14 269

13.10 Over- and under-representation of links to Kinh farmers comparedto the commune average, 2008–14 271

13.11 Differences in monthly per capita food expenditures in real1000 VND within minorities by (a) region, (b) ethnicity, and(c) language, 2008–14 272

xix

List of Figures

List of Tables

1.1 Sectoral distribution of production in selected Southeast Asian countries 5

1.2 General government net lending (% of GDP)(averages over the period) 10

1.3 General government gross debt (% of GDP) (averages over the period) 10

1.4 Mortality rate, under-5, female and male 17

1.5 Tertiary school enrolment in 2006 and 2011, females and males 19

1.6 Poverty headcount ratio at US$1.25 a day (PPP) (% of population) 20

2.1 Sample size and survey months 27

2.2 Households and communes (balanced panel) 32

2.3 Extent of attrition and comparison of attrition and non-attritionhouseholds 40

2.4 Comparison of gender, age, ethnicity, and illiteracy forhousehold (hh) heads 46

3.1 Communes in commune panel sample by 2014 income tertile and region 52

4.1 Proportion of households growing rice, by year, province, and quintile 75

4.2 Proportion of rice-growing households who sell, by province and year 76

4.3 Average proportion of rice output sold, by location, quintile, and year 77

4.4 Proportion of households growing one or more cash crop, by province,quintile, and year 78

4.5 Proportion of households engaged in aquaculture activity on theirown land, by province, quintile, and year 78

4.6 Proportion of households engaged in some commercial agriculturalactivities in all years of the panel 79

4.7 Characteristics of households engaged in selling rice, comparedto non-sellers 81

4.8 Regression results for correlates of selling rice and growing cash crops 83

4.9 Regression results for correlates of engagement in aquaculture 85

5.1 Economic activities of households, 2008–14 95

5.2 Proportion of income earned from different economic activities, 2008–14 95

5.3 Enterprise characteristics 96

5.4 External employment descriptive statistics 97

5.5 Economic activity transition matrices, 2008–14 99

5.6 Welfare measures, 2008–14 100

5.7 Impact of diversification on household welfare 103

5.8 Impact of diversification out of agriculture on household welfare 104

5.9 Determinants of the transition out of agriculture 106

5.A1 List of industry sectors of enterprise operation 108

5.A2 List of industry sectors of external employment 109

5.A3 Summary statistics 109

5.A4 Impact of diversification on household welfare, results for controlvariables 110

5.A5 Impact of diversification out of agriculture on household welfare,results for control variables 111

6.1 Property rights and agricultural investment, plot-level regressions 133

6.2 Property rights and agricultural investment, region-specific regressions 134

7.1 Inter-province and intra-province migration 142

7.2 Province of origin 143

7.3 Province of destination 143

7.4 Reasons for migrating 144

7.5 Distribution of households by migration status and foodexpenditure quintile 145

7.6 Household characteristics by migration status 145

7.7 Household characteristics by reason of migration 146

7.8 Migrant and working migrant characteristics 147

7.9 Migrant occupation 148

7.10 Role of migration networks 148

7.11 Remittance-recipient and non-remittance-recipient householdcharacteristics 149

7.12 Remittance use 150

7.13 Migration and food expenditure 152

7.14 Migration and natural shocks 153

7.15 Remittances and food expenditure 154

7.16 Migration, remittances, and borrowing behaviour 155

8.1 Mean comparison across households with and without a phone, 2014 165

8.2 Mean comparison across households with and without internetaccess, 2014 170

xxii

List of Tables

8.3 Dependent variable: number of phones 173

8.4 Dependent variable: internet access 175

9.1 Social capital and income, simple models 196

9.2 Social capital and income, comprehensive models 197

10.1 Summary properties of VARHS welfare measures 207

10.2 Composition of household income using summary income measure (%) 210

10.3 More detailed analysis of household income composition, 2008–14 (%) 211

10.4 Levels and changes in real per capita food consumption in the 2006–14VARHS panel 213

10.5 Percentage of households experiencing significant increases andreductions in food expenditure and income over the periodof the surveys 214

10.6 Extent and nature of attrition in the VARHS 2006–12 panel 215

10.7 Regression results for changes in welfare measures from 2006–14(with district-level fixed effects) 217

10.8 Regression results for changes in welfare measures within the VARHSpanel (with district-level fixed effects) 218

10.A1 Factor index weights for asset index 220

11.1 Characteristics of female-headed households, 2008–14 225

11.2 Household income and assets and female-headed households, 2008–14 225

11.3 Sources of income and female-headed households, 2008–14 226

11.4 Vulnerability of female-headed households, 2008–14 227

11.5 Gender cohort analysis 2008–14, health outcomes 229

11.6 Gender cohort analysis 2008–14, education outcomes 230

11.7 Gender cohort analysis 2008–14, economic activities 231

11.8 Indicators of female empowerment, 2008–14 232

11.9 Female empowerment and welfare, food consumption per capita 234

12.1 Geographical variation in fertility 240

12.2 Characteristics of households with children, 2008–14 241

12.3 Characteristics of different cohorts of children, 2008–14 243

12.4 Evolution of education outcomes for children aged 6–9 in 2008,by food expenditure quintile in 2008 245

12.5 Characteristics of different cohorts by gender of children, 2008–14 247

12.6 Characteristics of different cohorts by ethnicity of household head,2008–14 247

12.7 Panel data analysis of determinants of child welfare, 2008–14,6–18-year-olds 250

xxiii

List of Tables

12.8 Panel data analysis of determinants of child welfare, 2008–14,10–15-year-olds 252

12.9 Panel data analysis of female empowerment and child welfare,2008–14, 10–15-year-olds 253

13.1 Descriptive statistics by ethnicity for 2014 258

13.2 Income diversification by ethnicity in 2014 263

xxiv

List of Tables

List of Abbreviations

ADSL asymmetric digital subscriber line

AERC African Economic Research Consortium

ARDSPS Agriculture and Rural Development Sector Programme Support

BSPS Business Sector Programme Support

CAP-IPSARD Centre for Agricultural Policy Consulting of the Institute of Policy andStrategy for Agriculture and Rural Development

CBMS Community-Based Monitoring System

CEMA Committee for Ethnic Minority and Mountainous Area Affairs

CIEM Central Institute of Economic Management

CPI consumer price index

CPR common pool resources

DERG Development Economics Research Group

DHS Demographic and Health Surveys

EAs enumeration areas

FAO Food and Agricultural Organization

FDI foreign direct investment

GDP gross domestic product

GSO General Statistics Office

HLP High-Level Panel of Eminent Persons

ICT information and communication technology

ILO International Labour Organization

ILSSA Institute of Labour Science and Social Affairs

IMF International Monetary Fund

IPSARD Institute of Policy and Strategy for Agriculture and Rural Development

IRRI International Rice Research Institute

ISS Informal Sector Survey

IT information technology

KOICA Korea International Cooperation Agency

LFS Labour Force Survey

LUCs land-use certificates

MARD Ministry of Agriculture and Rural Development

MDG Millennium Development Goals

MO mass organization

MOLISA Ministry of Labour, Invalids, and Social Affairs

MPI Ministry of Planning and Investment

MRD Mekong River Delta

OLS ordinary least squares

PCI Provincial Competitiveness Index

SAM Social Accounting Matrix

SDG Sustainable Development Goals

SME Small and Medium-Scaled Enterprise

TIME Trinity Impact Evaluation Unit

UNU-WIDER United Nations University World Institute for DevelopmentEconomics Research

VARHS Viet Nam Access to Resources Household Survey

VBSP Bank for Social Policies

VEC Viet Nam Enterprise Census

VES Viet Nam Enterprise Survey

VHLSS Viet Nam Household Living Standards Survey

VLSS Viet Nam Living Standard Survey

VNPT Viet Nam Post and Telecommunications Group

WDI World Development Indicators

WTO World Trade Organization

WVS World Values Survey

xxvi

List of Abbreviations

Notes on Contributors

Ulrik Beck is a PhD student of economics at the University of Copenhagen, Denmark.He holds BA and MA degrees in Economics from University of Copenhagen and hasbeen a visiting graduate student at Cornell University and UC-Berkeley. His researchinterests are development economics using applied microeconomics with a focus onagricultural issues as well as measurement of poverty and inequality.

Kasper Brandt is a research assistant at the Department of Economics, University ofCopenhagen. He is enrolled in the Economics Master’s Programme at the University ofCopenhagen, and has applied for a PhD scholarship. His work is predominantly relatedto the microeconomics of development economics in Viet Nam and a few Africancountries.

Chiara Cazzuffi is a researcher at Rimisp—Latin American Centre for Rural Develop-ment, Santiago. Her research interests are in the area of development economics, agri-culture, andmigration. Sheholds a PhD in Economics from theUniversity of Sussex, andhas been involved in internationally-funded research projects on various aspects ofspatial inequality in Latin America and its impacts on poverty and development.

Heidi Kaila is a PhD student in economics at the University of Copenhagen. Herresearch interests lie within the area of both theoretical and empirical developmenteconomics with a focus on rural households. She holds a Bachelor’s and a Master’s inEconomics from the University of Helsinki, has worked as a research assistant at UNU-WIDER, and been a visiting PhD student at Cornell University.

Christina Kinghan is a PhD student in economics at Trinity College Dublin. Herresearch interests are in the area of micro, small, and medium-sized enterprise devel-opment in developing countries. She holds an MA in Economics from UniversityCollege Dublin and a Bachelor of Business Studies from Trinity College Dublin andhas been a PhD Intern at UNU-WIDER.

AndyMcKay is Professor of Development Economics at the University of Sussex, wherehe has worked since 2006. He works extensively on issues of poverty and dynamics ofliving standards, as well as labour, trade, agriculture and the distributional impact ofpolicy. While he has worked a lot on sub-Saharan Africa, he has worked on Viet Namsince 2003, but particularly since 2008 based on the VARHS. In addition he currentlyworks on projects on issues of female labour supply in Africa and Asia, and has beenManaging Editor of the Review of Development Economics since 2015.

Thomas Markussen is an Associate Professor at DERG, Department of Economics,University of Copenhagen. His research focuses on collective action, institutions and

the political economy of development and he has published in leading internationaljournals. His work on Viet Nammostly concerns landmarkets, land rights, and politicaleconomy.

Gaia Narciso is an Assistant Professor at the Department of Economics, Trinity CollegeDublin. Her areas of research are Development Economics, Migration and PoliticalEconomy. Her research has been widely cited and featured in international media.She has extensive experience in survey design and implementation and has conductedrandomized control trials in Ireland and abroad. She is one of the founders of theTrinity Impact Evaluation Unit (TIME).

Carol Newman is an Associate Professor at the Department of Economics, TrinityCollege Dublin. Her research is the microeconomics of development with a focus onhousehold and enterprise behaviour. She is involved in a number of major researchprojects in Africa and Southeast Asia and is an expert on economic development in VietNam. She has published widely in the fields of development economics and agriculturaleconomics.

Emilie Perge is an Adjunct Associate Research Scientist at the Agriculture and FoodSecurity Centre at the Earth Institute, Columbia University. Her research consists ofanalysing synergies between environment and economic development focusing onforest landscapes, fisheries, poverty, and household livelihoods. She is currently work-ing on tree-planting for climate adaptation in sub-Saharan Africa and on coping withshocks using forest products and areas. She holds a DPhil in Economics from theUniversity of Sussex and an MA in Environmental Economics from Université ParisX Nanterre.

Saurabh Singhal is a Research Fellow at UNU-WIDER. His research interests are thepolitical economy of development and microeconomic analysis of household andindividual decision-making in developing countries using observational and experi-mental data. He holds a PhD in Economics from the University of Southern California,and a MA in Economics from the Delhi School of Economics.

Finn Tarp is Director of UNU-WIDER and Coordinator of DERG at the University ofCopenhagen. He has some thirty-eight years of experience in academic and applieddevelopment economics research and teaching; and his field experience covers morethan twenty years of in-country work in thirty-five countries across Africa and thedeveloping world more generally, including longer-term assignments in Swaziland,Mozambique, Zimbabwe, and Viet Nam. He is a leading international expert on issuesof development strategy and foreign aid, with a sustained interest in poverty, incomedistribution and growth. He has published widely in leading international academicjournals alongside a series of books, and he is a member of the World Bank ChiefEconomist’s Council of Eminent Persons, and a long-time resource person of theAfrican Economic Research Consortium (AERC). He has been awarded the VietnameseGovernment Medals of Honour for Support to the Planning and Investment System,and the Cause of Science and Technology. In November 2015 Finn Tarp was knightedwhen Her Majesty Queen Margrethe of Denmark awarded him the Order of theDannebrog.

xxviii

Notes on Contributors

1

Viet Nam

Setting the Scene

Finn Tarp

1.1 Introduction

Viet Nam is a populous Southeast Asian economy with a particular socioeco-nomic and political history.1 At the end of the ‘American War’ in 1975 ambi-tions for the future were high; but despite great potential, the economyremained poor. International isolation played its role as did centralist policies;and the five-year plan adopted in 1976 turned out to be a complete failure.Economic policies started to be reversed following economic collapse in themid-1980s, and Viet Nam initiated its home-grown Doi Moi (renovation) reformprocess in 1986. Accordingly, wide-ranging institutional reforms have beengradually implemented in the country since then, including a greater relianceon market forces in the allocation of resources and the determination of prices.A shift from an economy dominated by the state and cooperative sectors to asituationwhere the private sector and foreign investment account for a relativelyhigh proportion of gross domestic product (GDP) can also be noted.

Some of the policy measures taken in Viet Nam over the past three decadeswithin the framework of the Doi Moi process may seem to correspond withthe tenets of the ‘orthodox’ International Monetary Fund (IMF)- and WorldBank-supported stabilization and structural programmes pursued in the 1980sand 1990s throughout the developing world (see Tarp 1993). Yet there havealso been many significant differences. First of all, Viet Nam has aspiredexplicitly at furthering a transition from a centrally planned to a socialist

1 See, for example, <https://en.wikipedia.org/wiki/History_of_Vietnam_since_1945> (accessed8 May 2016).

https://en.wikipedia.org/wiki/History_of_Vietnam_since_1945

market economy, rather than pursuing outright liberalization of the domesticeconomy and international transactions.2 The role of the Communist Partyand the state apparatus has, therefore, continued to be prominent in allaspects of socioeconomic and political life (Newman, van den Broeck, andTarp 2014). To illustrate, the government has continued to intervene inagricultural markets (Markussen, van den Broeck, and Tarp 2011), and avigorous industrial policy, moving only step by step from import-substitutionto export-promotion has been a key attribute (Abbott and Tarp 2012). Simi-larly, Doi Moi did not focus immediate attention on international tradeliberalization and World Trade Organization (WTO) membership. Instead,centrally coordinated public investments, targeted policies, and institutionalinitiatives were put in place, as outlined by Abbott, Bentzen, and Tarp (2009),much along the lines pursued in other East Asian countries in the 1970s as partof their export push strategies. WTO membership came, in fact, rather late inthe process for Viet Nam, less than ten years ago, in 2007.Consequent socioeconomic outcomes have been impressive, and a central

argument throughout the chapters of this volume is that the developmentcommunity has a lot to learn fromViet Namwhen it comes to the formulationand implementation of economic reform and effective development policy.With regard to this background, this introductory chapter aims, first, to

provide the reader with an overview of how Viet Nam’s general economic andsocioeconomic performance and characteristics have evolved over the past fewdecades, basedon standard data available from international sources such as theWorld Development Indicators (WDI) of theWorld Bank. To add internationalcomparative perspective, Viet Nam is, in this chapter, matched throughoutSections 1.1 to 1.3 with a group of regional counterparts, commonly China,Thailand, Indonesia, and Cambodia.3 The general aim is to set the scene for theremaining chapters where focus is on the household sector in the rural areas ofViet Nam. Section 1.4 connects the macro-setting referred to as well as back-ground material regarding the panel data set relied on in subsequent chapters,moving on in Section 1.5 to the underlying questionnaires implemented in theViet Nam Access to Resources Household Survey (VARHS) carried out everysecond year from2006 onwards. Section1.6 outlines the remainder of thebook.

1.1.1 Macroeconomic and Monetary Performance

Associated with the successful implementation of the Doi Moi reform pro-gramme, Viet Nam has in many ways been among the most successful East

2 See <http://vietnamlawmagazine.vn/new-concept-of-socialist-oriented-market-economy-introduced-4582.html> (accessed 8 May 2016).

3 This group of countries is regularly referred to in policy debates in Viet Nam, and is clearly theset of countries Vietnamese policy makers typically have in mind when they look for comparators.

Viet Nam: Setting the Scene

2

http://vietnamlawmagazine.vn/new-concept-of-socialist-oriented-market-economy-introduced-4582.html

http://vietnamlawmagazine.vn/new-concept-of-socialist-oriented-market-economy-introduced-4582.html

Asian economies. This is certainly so in terms of GDP growth. Yet, progress hasby nomeans been linear, and Viet Nam, which acquired lower middle-incomestatus in 2010, continues to be relatively poor in regional comparison.Figures 1.1, 1.2, and 1.3 illustrate these points vividly.

Figure 1.1 shows, first of all, the excellent performance of the economy inthe 1990s. It also demonstrates the significant impact of the Asian financialcrisis in 1997, which was a major economic blow. The 2007–8 global financialcrisis hadmuch less impact, in largemeasure due to Viet Nam’s effectivemacro-economic response; inmore recent years, the annual growth ratehas returned toa level above 6 per cent.4

Turning to Figure 1.2, it is evident that Viet Nam has been outperformed byChina in terms of GDP growth—as has the rest of the world. It is equally clearthat Viet Nam has done much better—and has had a much more stableperformance—than Southeast Asian countries such as Thailand, Indonesia,and Cambodia. The former two were particularly badly hit by the Asianfinancial crisis and have suffered large economic fluctuations, in contrast toViet Nam.

There should be no room for complacency though. Figure 1.3 confirms thefact that Viet Nam remains a relatively poor country, with a GDP/capita wellbelow that of Malaysia, China, and Thailand and closer to the Philippines, yetabove that of Cambodia and Laos.

The solid aggregate economic growth in Viet Nam has, over the years, beenassociated with what most observers would characterize as a process of

0

2

4

6

8

10

12

1985 1989 1993 1997 2001 2005 2009 2013

Perc

enta

ge

po

ints

Year

Figure 1.1 Real GDP growth, Viet NamSource: World Bank World Development Indicators.

4 See Thurlow et al. (2011), and note that, while not fully captured in Figure 1.1, growth wouldappear to have picked up since 2013, approximating an average of 6.5 per cent per year for the2000–15 period.


3

successful structural transformation, involving sectoral reallocation from agri-culture to higher productivity sectors. Figure 1.4 and Table 1.1 demonstratethis point. The long-run sectoral trends of agriculture, industry, and services inFigure 1.4 are impressive even if there seems to be some tapering off in more

–15

–10

–5

0

5

10

15

1985 1989 1993 1997 2001 2005 2009 2013

Perc

enta

ge

po

ints

Year

Cambodia China Indonesia Thailand Vietnam

Figure 1.2 Real GDP per capita growth in selected countries, 1985–2013Source: World Bank World Development Indicators.

0

1000

2000

3000

4000

5000

6000

7000

8000

Cambodia China Indonesia Lao PDR Malaysia Philippines Thailand Vietnam

Co

nst

ant

2005

US

$

Country

2000 2013

Figure 1.3 Real GDP per capita in Southeast Asian countriesSource: World Bank World Development Indicators.


4

recent years. Table 1.1 underpins this, by comparing the country to a selectedgroup of Southeast Asian counterparts.

One concern that merits mentioning is that value added per worker in theagriculture sector according to the WDI data base (measured in constant 2005US dollars (US$)) only grew marginally from 2006 to 2013 in Viet Nam assuggested in Figure 1.5. China, Indonesia, and Thailand all appear to havedone better at significantly higher levels as well. While Indonesia’s agricul-tural output per worker also stagnated in the last decade, it, nevertheless, hasremained above that of Viet Nam.5

15

20

25

30

35

40

45

50

1985 1989 1993 1997 2001 2005 2009 2013

Perc

enta

ge

of

GD

P

Year

Agriculture Industry Services

Figure 1.4 Sectorial distribution of aggregate Vietnamese outputSource: World Bank World Development Indicators.

Table 1.1 Sectoral distribution of production in selected Southeast Asian countries

1985 1995 2005 2013

Cambodia Agriculture n/a 49.6 32.4 33.5Industry n/a 14.8 26.4 25.6

China Agriculture 28.4 20.0 12.1 10.0Industry 42.9 47.2 47.4 43.9

Indonesia Agriculture 23.2 17.1 13.1 14.4Industry 35.8 41.8 46.5 45.7

Thailand Agriculture 15.8 9.5 10.3 12.0Industry 31.8 40.7 44.0 42.5

Viet Nam Agriculture 40.2 27.2 19.3 18.4Industry 27.4 28.8 38.1 38.3

Note: Percentage share of agricultural and industrial production of total GDP. Services is the residual category so its shareis 100% minus the share of agriculture and industry.

Source: World Bank World Development Indicators.

5 The topic of agricultural productivity is pursued further in Chapter 4.


5

Turning briefly to technology infrastructure in the form of fixed broadbandsubscriptions per hundred persons, these grew from 0.6 to 5.6 per centbetween 2006 and 2013. Viet Nam still lags behind China and Thailand,while Indonesia and Cambodia are even further behind, as shown inFigure 1.6.Figure 1.7 highlights the fact that Viet Nam has benefited from a significant

demographic dividend, which is high even by Southeast Asian comparison.The share of the 15–64-year-old population has increased from 55 per cent in

0

200

400

600

800

1000

1200

2006 2007 2008 2009 2010 2011 2012 2013 2014

Co

nst

ant

2005

US$

Year


Figure 1.5 Agriculture value added per worker (constant 2005 US$)Source: World Bank World Development Indicators.

0

2

4

6

8

10

12

14

16

2006 2007 2008 2009 2010 2011 2012 2013

Sub

scri

pti

on

s 10

0 p

eop

le

Year


Figure 1.6 Fixed (wired) broadband subscriptions (per 100 people), 2006–13Source: World Bank World Development Indicators.


6

the mid-1980s to over 70 per cent in recent years, almost on par with theshares in China and Thailand. Labour-force participation rates are also high(see Figure 1.8). The International Labour Organization (ILO) estimates of thelabour-force participation of the 15–64-year-old population has remainedbetween 81 and 85 per cent since the 1990s, slightly higher than in Chinaand Thailand for most of the same period.

Turning to the monetary sector of the economy, Viet Nam has, over theyears, had its share of high inflation experiences, most dramatically in themiddle of the economic collapse in the mid-1980s. Then inflation, as meas-ured by the annual increases in the consumer price index (CPI), exploded tomore than 450 per cent and only gradually came down to more modest levelsfrom the mid-1990s onwards.

Figure 1.9 illustrates that domestic prices were also significantly affected bythe 2007–8 and 2011 price spikes in international food prices, before droppingdown to about 5 per cent on an annual basis, pretty much in line with theGDP growth rate as discussed. With regard to the Southeast Asian perspective,CPI inflation in Viet Nam was relatively high from 2007 onwards, but from2012 it has been more in line with the experience of other countries in theregion (see Figure 1.10). The monetary policy interest rate remains high,though, as shown in Figure 1.11.

45

50

55

60

65

70

75

1985 1989 1993 1997 2001 2005 2009 2013

Perc

enta

ge

of

tota

l po

pul

atio

n

Year


Figure 1.7 Share of population aged 15–64 years (% of total population)Source: World Bank World Development Indicators.


7

-5

0

5

10

15

20

25

1994 1998 2002 2006 2010 2014

Perc

enta

ge

chan

ge

per

an

num

Year

Figure 1.9 Inflation in Viet Nam (annual changes in % in the CPI)Source: IMF World Economic Outlook.

55

60

65

70

75

80

85

1991 1995 1999 2003 2007 2011

Perc

enta

ge

of

tota

l po

pul

atio

n

Year


Figure 1.8 Employment of 15-year-olds and older to total population (in %)Source: World Bank World Development Indicators.


8

Finally, reflecting the somewhat more expansionary macroeconomic policyline Viet Nam adopted after the turn of the century, which has underpinnedthe growth performance, government borrowing has gradually edgedupwards, as shown in Table 1.2. While government debt is higher than in

-5

5

15

25

35

45

55

65

1995 1999 2003 2007 2011

Perc

enta

ge

chan

ge

per

an

num

Year


Figure 1.10 Inflation in selected countries (annual changes in % in the CPI)Source: IMF World Economic Outlook.

0

2

4

6

8

10

12

14

16

18

20

1996Q1 1998Q1 2000Q1 2002Q1 2004Q1 2006Q1 2008Q1 2010Q1 2012Q1 2014Q1

Perc

enta

ge

Quarter

Figure 1.11 Monetary policy interest rate (Viet Nam)Source: IMF International Financial Statistics.


9

other Southeast Asian countries there would appear to be no major reason forconcern at present on this account, as the government gross debt ratio as ashare of GDP is only slightly above 50 per cent, as indicated in Table 1.3.In addition, the external macroeconomic performance to which I turn in

Section 1.2 is very convincing. It can also be noted here that while domesticcredit provided by the banking sector (as a share of GDP) grew substantially inViet Nam during 2006–13, it has nevertheless declined as a share of GDP since2010 (see Figure 1.12). A roughly similar development can be seen inFigure 1.13 for domestic credit to the private sector during 2006–13, puttingViet Nam below China and Thailand, and above Indonesia and Cambodia.

1.2 External Economic Relations

Viet Nam’s international economic performance has been strong for manyyears and the country is a very open economy as measured by standardindicators. Trade as a share of GDP has increased steadily for over fifteenyears and is by now higher than that of Thailand (reflecting in part the factthat Viet Nam weathered the 2007–8 crisis much better than Thailand).Furthermore, while China and Viet Nam both started with a trade/GDP ratioof about 20 per cent in 1986, the trade share of Viet Nam was, in 2013, muchhigher than that of China and Indonesia, as shown in Figure 1.14. Moreover,

Table 1.2 General government net lending (% of GDP) (averages overthe period)

1995–9 2000–4 2005–9 2010–14

Cambodia �5.1 �5.2 �1.0 �2.7China �1.1 �2.5 �0.9 �0.6Indonesia �0.5 �1.1 �0.3 �1.5Thailand �2.2 �1.4 0.2 �1.0Viet Nam �0.9 �2.1 �1.9 �4.4

Source: IMF World Economic Outlook.

Table 1.3 General government gross debt (% of GDP) (averages overthe period)

2000–4 2005–9 2010–14

Cambodia 38.9 31.2 29.0China 37.1 33.5 38.2Indonesia 66.1 33.5 24.1Thailand 54.1 42.0 44.6Viet Nam 36.1 40.7 50.9

Source: IMF World Economic Outlook.


10

0

20

40

60

80

100

120

140

160

180

2006 2007 2008 2009 2010 2011 2012 2013

Perc

enta

ge

of

GPD

Year


Figure 1.13 Domestic credit to private sector (% of GDP)Source: World Bank World Development Indicators.

0

20

40

60

80

100

120

140

160

180

200

2006 2007 2008 2009 2010 2011 2012 2013

Perc

enta

ge

of

GD

P

Year

Cambodia China Indonesia Malaysia Thailand Vietnam

Figure 1.12 Domestic credit provided by financial sector (% of GDP)Source: World Bank World Development Indicators.


11

while the trade balance fluctuated below zero until around 2007 the trendchanged in that year and the balance turned positive in 2011 (see Figure 1.15).A similar development can be observed in the current account balance,

which improved significantly from around 2007. Viet Nam is by now in astronger relative position than any of the other countries included inFigure 1.16. The strong external position of Viet Nam is equally clear fromthe foreign direct investment (FDI) net inflows. Viet Nam has attracted sub-stantial amounts of foreign investment over the past twenty-five years. In fact,Viet Nam is, in this regard, a star performer throughout the period from thelate 1980s, as shown in Figure 1.17, where FDI inflows to Viet Nam as a ratioof GDP have consistently been higher than to China and Thailand. OnlyCambodia is on par with Viet Nam, as measured by this indicator, whileIndonesia trails far behind.6

While total international reserves have dropped somewhat since 2007, andare relatively low in Viet Nam (see Figure 1.18) there would appear to be littlereason for concern. This is also reflected in the downward sloping, yet verystable exchange rate development vis-à-vis the US dollar after the massive

0

20

40

60

80

100

120

140

160

180

1986 1990 1994 1998 2002 2006 2010 2014

Perc

enta

ge

of

GD

P

Year


Figure 1.14 Trade (exports plus imports) as a share of GDP (%)Source: World Bank World Development Indicators.

6 See Hansen, Rand, and Tarp (2003) for an early overview of where this FDI has come from andinto what sectors it has gone.


12

–15

–10

–5

0

5

10

15

1992 1996 2000 2004 2008 2012

Perc

enta

ge

of

GD

P

YearCambodia China Indonesia Thailand Vietnam

Figure 1.16 Current account balance as a share of GDP (%)Source: IMF World Economic Outlook.

–20

–15

–10

–5

0

5

10

15

20

1986 1989 1992 1995 1998 2001 2004 2007 2010 2013

Perc

enta

ge

of

GD

P

YearCambodia China Indonesia Thailand Vietnam

Figure 1.15 Trade balance as a share of GDP (%)Source: World Bank World Development Indicators.


13

0

10

20

30

40

50

60

1995 1999 2003 2007 2011

Perc

enta

ge

of

GD

P

Year


Figure 1.18 Total reserves excluding gold as a share of GDP (%)Source: World Bank World Development Indicators.

–4

–2

0

2

4

6

8

10

12

14

1985 1989 1993 1997 2001 2005 2009 2013

Perc

enta

ge

of

GD

P

Year


Figure 1.17 Foreign direct investment, net inflows (% of GDP)Source: World Bank World Development Indicators.


14

external adjustments in 1986–8 (see Figure 1.19). In an international com-parative perspective this performance is quite impressive.

1.3 Household Consumption and Socioeconomic Indicators

The significant economic growth in GDP in Viet Nam has been accompaniedby growth in household final consumption at about the same rate. This isshown in Figure 1.20, which reflects an average annual rate of increase of 6 percent in household consumption from 2006–13. To compare, householdconsumption in Thailand only grew by 2.5 per cent per year in this periodwhereas consumption in most years grew faster in China, especially from2009, with the exception of 2010.

A truly impressive socioeconomic characteristic of Viet Nam is life expect-ancy at birth. Female life expectancy has consistently outperformed all com-parator countries, as shown in Figure 1.21, and is remarkably high (eightyyears on average) between 2006 and 2013. This is almost ten years more thanIndonesia and on par with many developed countries. Male life expectancy isalso relatively high (see Figure 1.22), on par with Thailand and better thanIndonesia and Cambodia, though trailing behind China. This performancereflects sustained investment in public health care over many years in Viet

1

10

100

1000

10000

100000

1986

Q1

1988

Q1

1990

Q1

1992

Q1

1994

Q1

1996

Q1

1998

Q1

2000

Q1

2002

Q1

2004

Q1

2006

Q1

2008

Q1

2010

Q1

2012

Q1

2014

Q1

Year

Ind

ex v

alue

, 199

0 Q

1 =

100,

log

sca

le

Figure 1.19 US dollar/Vietnamese Dong exchange rateNote: Quarterly data with period averages. Index value with 1990Q1=100. Lower value indicatesdepreciation of the Dong.Source: IMF International Financial Statistics.


15

Nam as well as dietary habits, which may only now be changing somewhatwith the introduction of Western food habits and consumption items.In terms of under-5 mortality rates Viet Nam performs much less convin-

cingly, as is clear from Table 1.4. While significant progress has been madesince 2000 for both girls and boys, Viet Nam only occupies a middle ground—better than Indonesia and Cambodia and below the performance of Chinaand Thailand.

69

71

73

75

77

79

81

2006 2007 2008 2009 2010 2011 2012 2013

Year

s

Year

Indonesia Cambodia China Malaysia Thailand Vietnam

Figure 1.21 Life expectancy at birth, female (years), 2006–13Source: World Bank World Development Indicators.

–2

0

2

4

6

8

10

12

14

2006 2007 2008 2009 2010 2011 2012 2013

An

nua

l per

cen

tag

e g

row

th

Year


Figure 1.20 Household final consumption expenditure (% annual growth)Source: World Bank World Development Indicators.


16

The prevalence of undernourishment as a share of the Vietnamese popula-tion dropped from close to 19 per cent in 2006 to 13 per cent in 2013, asshown in Figure 1.23, more or less in line with the general drop in the povertyheadcount rate discussed in Section 1.4.

While this improvement is clearly better than the experience of Cambodia,it is not significantly different from what was seen in China and Thailand.This means that Viet Nam has relatively fewer undernourished people thanCambodia, more than China and Indonesia, andmanymore than Thailand inparticular. A roughly similar picture emerges when focus is on the food deficit,as in Figure 1.24, reflecting Indonesia’s relative effectiveness in eradicatingmalnourishment and the food deficit during 2006–13.

Turning finally to the level of education, there is not much differencebetween the countries in focus in this chapter when it comes to primary and

65

66

67

68

69

70

71

72

73

74

75

2006 2007 2008 2009 2010 2011 2012 2013

Year

s

Year

Indonesia Cambodia China Malaysia Thailand Vietnam

Figure 1.22 Life expectancy at birth, male (years), 2006–13Source: World Bank World Development Indicators.

Table 1.4 Mortality rate, under-5, female and male

Mortality rate, under-5, female (per 1,000) Mortality rate, under-5, male (per 1,000)

2000 2010 2013 2000 2010 2013

China 34.7 14.7 11.8 38.9 16.9 13.5Thailand 19.2 12.5 11.3 25.6 16.3 14.7Viet Nam 30.3 22.3 20.5 39.6 29.4 26.9Indonesia 46.9 29.1 25.6 57.2 37.1 32.9Cambodia 102.5 38.9 33.5 118.1 48.4 42.2



17

45

55

65

75

85

95

105

115

125

135

145

2006 2007 2008 2009 2010 2011 2012 2013 2014

Kilo

calo

ries

per

per

son

per

day

Year


Figure 1.24 Depth of the food deficit (kilocalories per person per day)Note: The depth of the food deficit indicates how many calories would be needed to lift theundernourished from their status, everything else being constant. The average intensity of fooddeprivation of the undernourished, estimated as the difference between the average dietary energyrequirement and the average dietary energy consumption of the undernourished population (food-deprived), is multiplied by the number of undernourished to provide an estimate of the total fooddeficit in the country, which is then normalized by the total population.Source: World Bank World Development Indicators.

5

7

9

11

13

15

17

19

21

2006 2007 2008 2009 2010 2011 2012 2013

Perc

enta

ge

of

po

pul

atio

n

Year


Figure 1.23 Prevalence of undernourishment (% of population)Source: World Bank World Development Indicators.

18

secondary education. However, when focus is on university education Thai-land is far ahead of both Viet Nam and the other countries in the comparisongroup (Table 1.5). The share of tertiary school enrolment grew by about 8percentage points in Viet Nam from 2006 to 2011 for both females andmales, at a much lower level than Thailand. In comparison to the othercountries, the only noticeable difference is that Cambodia stands out asbeing very far behind.

1.4 Moving from the Macro to the Household Level

The general macroeconomic and socioeconomic framework reviewed inSections 1.1 and 1.2, together with the ongoing institutional reforms, setsthe general scene within which developments at the household level in ruralareas of Viet Nam have evolved over the past decades; there is no doubt thataggregate progress has, in general, trickled down to poor people. This isreflected in Figure 1.25 and Table 1.6, which provide a comparison (basedon the World Bank poverty line of US$1.25 (PPP)), with regional counterpartcountries, of the development of their poverty headcount ratios.

Comparison of poverty rates has to be taken very cautiously given theinherent data issues. Nevertheless, Viet Nam stands out remarkably by thismeasure. While data are not available for Viet Nam in the 1980s, widespreadprogress can be seen in all countries included. Moreover, and relevant forpresent purposes, Viet Nam started out with the highest poverty rate in the1990s (57 per cent), and in 2010–12 had the lowest poverty rate (3 per cent),except for Thailand (and Malaysia). Yet, these two countries had povertyrates of 0.2 and 2 per cent respectively in the 2000s when Viet Nam wasstill at 27 per cent. One can also compare to, for example, Indonesia andthe Philippines, which from 2010–12 had poverty rates of 17–19 per cent

Table 1.5 Tertiary school enrolment in 2006 and 2011, females and males

School enrolment, tertiary,female (% gross)

School enrolment, tertiary,male (% gross)

2006 2011 2006 2011

China 19.0 25.7 20.0 23.1Thailand 45.8 58.8 42.6 46.4Viet Nam 16.1 24.6 16.8 24.2Indonesia 17.0 25.0 18.8 29.4Cambodia 3.6 12.0 7.6 19.6



19

even if they are at much higher real GDP/capita levels (see Figure 1.3) thanViet Nam.At the same time, while this general picture and the underlying trend are

illustrative and encouraging, as well as in line with the biannual nationallyrepresentative Viet Nam Household Living Standard Survey (VHLSS) carried

0

10

20

30

40

50

60

70

80

China Indonesia Cambodia Lao PDR Malaysia Philippines Thailand Vietnam

Perc

enta

ge

of

po

pul

atio

n

Country

1980s 1990s 2000s 2010–2012

Figure 1.25 Poverty headcount ratio at US$1.25 a day (PPP) (% of population)Note: The poverty headcount ratio is based onWorld Bank estimates and obtained from the WorldBank Development Indicators database. The estimates are not available for consecutive years butmeasured on average every two years for each country and not for the same countries in the sameyears. Therefore, the time categories are based on simple averages except for the 1980s when thereis only one observation per country excluding the Philippines that has two estimates in 1985 and1988. There is missing information for Cambodia, Lao PDR, and Viet Nam for the 1980s, andMalaysia for 2010–12.Source: World Bank World Development Indicators.

Table 1.6 Poverty headcount ratio at US$1.25 a day (PPP) (% ofpopulation)

1980s 1990s 2000s 2010–12

China 54 47 19 8Indonesia 68 50 25 17Cambodia . . . 45 24 11Lao PDR . . . 52 38 30Malaysia 2 1 0.2 . . .Philippines 36 29 22 19Thailand 17 5 2 0.3Viet Nam . . . 57 27 3

Note: As in Figure 1.15.



20

out by the General Statistical Office (GSO),7 it does not provide insights intoa host of the many other policy-relevant issues which Vietnamese policymakers face.

The basic idea behind the original pilot of VARHS, carried out in 2002, wasthat existing surveys (including the VHLSS) did not provide the data andinformation needed for coming to grips with a series of intricate and pressingissues related to land, credit, and labour. Only scant information was, forexample, available on the way in which households in rural areas accessresources in these markets. This lack of knowledge appeared as a particularlycritical constraint to evidence-based policy-making. After all, Viet Nam had, in1986, embarked on a gradual process of liberating and transforming its econ-omy from a centrally planned command-type system to a more market-oriented-based allocation of resources. In such a context, the appropriatedevelopment of market institutions is an essential challenge.

VARHS set out to help fill this information gap, and this rationale remainedunchanged as the survey was implemented in the following decade. Forexample, making land and credit markets more efficient today is no less keyto sustaining private sector development than in 2002. VARHSwas alsomeantto help in better understanding the role land markets play in the allocation ofresources within the agricultural sector, including the possible influence oftenure security. Similarly, it was agreed from the very beginning that it wasnecessary to dig deep into the extent of land market transactions and whetherland rental or land sale transactions were active. Other land issues relate to, forexample, the impact of contract terms on efficiency and equity.

Another illustration of the need for additional data and information con-cerns the functioning of rural credit markets and the extent to which creditrationing impedes agricultural development. Further insights into these issues(with a view to improved policy-making) presume, first of all, availability ofdata on the amounts of credit that farmers have actually taken. In addition,data are needed on the investment projects they could not undertake for lackof credit facilities and on the consumption expenditures they could notfinance. If consumption credit is not readily available under distress condi-tions, it is evident that farmers will have to resort to costly alternative survivalstrategies such as sale of productive assets. And, if credit markets do not workproperly, farmers will not be able to repurchase their lost assets later, therebypotentially driving them into chronic poverty, suggesting that imperfectcredit markets may have serious impacts on consumption and humanwelfare.There are, in other words, interrelated issues of market development, of insti-tutions, and of poverty, which, it was agreed, merit attention.

7 While the level of the poverty rate based on VHLSS data is higher than indicated by theUS$1.25 (PPP) per day poverty line the underlying trend over time is about the same.


21

As a third example, it was accepted in the VARHS design process that there isa continuing need to help bring out data and information on issues related tothe fragmentation of the land. For this to be possible, it is, however, necessaryto collect data at individual plot level. The VARHS was specifically designed toillicit such information, providing a basis for a much more detailed under-standing of agricultural production than so far possible. It was also establishedthat this understanding should be extended to cover cross-cutting issues suchas the role of gender and poverty in labour market participation, agriculturalproduction and marketing, access to credit, risk, and to information; the database was designed so as to allow for the exploration of further issues related tothe role of ethnicity and, eventually, to a variety of other topics such aspolitical connections, migration, and happiness. The core issues of VARHShave, however, remained the same throughout, and while the associatedquestionnaires have, over the years, been developed and refined, the basicstructure and content have been prepared, always ensuring comparability, tobe able to exploit the panel nature of the data to the maximum.Accordingly, and to sum up this section, the purpose of the VARHS survey

has, throughout, been to deepen our understanding of household access, andlack of access, to productive resources in rural Viet Nam. The intention hasbeen to come to grips with why some households have restricted access toresources, and how these restrictions affect the household economy andwelfare. ‘Productive resources’ have been broadly defined to include physical,financial, human, and social capital, as well as land, and the survey hascollected information on a broad range of topics, such as rural employment,on- and off-farm income-generating activities, rural enterprises, propertyrights, savings, investment, insurance, and participation in formal and infor-mal social networks.Importantly, given that the same set of households was interviewed over

the years, as further discussed in Chapter 2, such detailed data allows for anilluminating analysis of structural change and its impacts at the micro levels.

1.5 The VARHS Questionnaires

The VARHS survey instrument used in all years included both a commune anda household questionnaire. The following types of detailed information werecollected, with minor modifications as the process went on. For example, the2012 survey introduced new sections on migration and remittances, socialproblems, happiness, and constraints to the expansion of household enter-prises. The questionnaires from specific years can be downloaded from theCentral Institute of Economic Management (CIEM) website: <http://www.ciem.org.vn/>.


22

http://www.ciem.org.vn/

http://www.ciem.org.vn/

A. COMMUNE QUESTIONNAIREInformation on interviewees:

Section 1: Demographic information and general information on thecommune

Section 2: MigrationSection 3: Development programmesSection 4: Agriculture: crops cultivated, land sales, land rental agreements,

types, and amount of landSection 5: Income and employment: main sources of income/employment,

and enterprise activitySection 6: Infrastructure: roads, waterways, electricity, markets, and schoolsSection 7: ShocksSection 8: Irrigation management: public/cooperative irrigation facilitiesSection 9: Credit and savings: possibilities for credit and saving: banks,

funds, unions, moneylendersSection 10: Commune problemsSection 11: Access to servicesSection 12: General information on interviewed persons

B. MAIN HOUSEHOLD QUESTIONNAIRECover page: Surveyor, date, and ethnicity/language:

Section 1: Household roster, general characteristics of household membersand housing

Section 2: Agricultural land (plot level) (including information on disasters)Section 3: Crop agricultureSection 4: Livestock, forestry, aquaculture, agricultural services, access to

markets, and common property resourcesSection 5: Employment, occupation, time use, and other sources of incomeSection 6: Extension servicesSection 7: Food expenditures, other expenses, savings, household durable

goodsSection 8: CreditSection 9: Shocks and risk copingSection 10: Social capital and networksSection 11: MigrationSection 12: Trust, political connections, sources of information, and rural

society

1.6 Book Outline

The structure of the present volume consists of four main parts, in addition tothe present scene-setting Chapter 1 and the succeeding Chapter 2, which


23

explore the characteristics of the VARHS data base. They focus on: (i) theongoing transformation in the rural economy of Viet Nam; (ii) key productionfactors and institutions; (iii) welfare and distributional issues; and (iv) lessonsand policy.Part I, on the transformation of the rural sector, is presented and analysed

from three different, complementary perspectives. They include a localcommune-level analysis (Chapter 3) in addition to two chapters on, respect-ively, the agriculture sector (addressing issues such as diversification, commer-cialization, and transformation) (Chapter 4) and the non-farm rural economy(Chapter 5). Part II first reviews the land and land markets (Chapter 6) andthen discusses labour and migration (Chapter 7), before digging into the roleof technology and innovation (Chapter 8), and finishing by addressing thecomplex issue of social capital and political connections (Chapter 9). Part III isconcerned with the critically important topics of welfare impacts and distri-butional issues. To begin, a rural household-level perspective is adopted toassess who are the winners and losers of economic development in Viet Nam(Chapter 10). Three chapters on gender (Chapter 11), children and youths(Chapter 12), and ethnicity (Chapter 13) complete Part III, while Part IV(Chapter 14) addresses lessons and policy.As far as is feasible, a common structure has been followed in the eleven

chapters that make up Parts I, II, and III. The authors proceed, first, topresent descriptive statistics to describe the main observations and correl-ations of interest using statistical tests to check for differences between thevariables of interest. Reference is made to other literature as appropriate,followed whenever relevant by regression analysis that allows the key cor-relations to be identified once controls (including, for example, householdfixed effects) are included. One exception to this is the chapter on ethnicity.Given that ethnicity is referenced in almost all other chapters, the authors ofthis chapter have not repeated the empirical models presented elsewhere inthe book.Statistics are presented, in general, by individual province when spatial

comparisons are made. This is not, however, always possible/relevant andsometimes a grouping of provinces has been preferred, as discussed furtherin Chapter 2, as an easier and more communicative way to present the data. Itis noted that while the VARHS is representative at provincial level the samecannot be said at national level. Yet VARHS does have an attractive spread andcomposition of provinces that covers the whole country, and, importantly,the VARHS 2006–14 panel covers the same 2,162 households (from 12 prov-inces and 466 communes) throughout the period and therefore provides aunique opportunity to capture what happened to them.To ensure consistency in welfare indicators across chapters, food expend-

iture and household income are provided in real terms on a monthly per


24

capita basis. The current food expenditure variable has been deflated by thenational food price index and inflated to reflect a standardmonth of 30.4 days(the questionnaire asks about expenditures in the last four weeks, i.e. twenty-eight days). Household income, which was collected in annual nominalfigures, was deflated by the national CPI and was further divided by twelveto get from yearly to monthly figures.

The final Part IV sums up key findings, addresses their policy implications,and discusses a number of wider perspectives, including a variety of pointsreferred to in Chapter 1 and Chapter 2, such as the impact of the internationaleconomic crises, which Viet Nam has managed relatively well.

References

Abbott, P. and F. Tarp (2012). ‘Globalization Crises, Trade and Development inVietnam’. Journal of International Commerce, Economics and Policy, 3(1): 1–23.

Abbott, P., J. Bentzen, and F. Tarp (2009). ‘Trade and Development: Lessons fromVietnam’s Past Trade Agreements’. World Development, 37(2): 341–53.

Hansen, H., J. Rand, and F. Tarp (2003). ‘Are FDI Inflows Complements or Substitutesacross Borders: Empirical Evidence from Five Asian Countries’. Paper prepared for theinternational workshop on Understanding FDI-Assisted Economic Developmentheld at the TIK centre at the University of Oslo, 22–25 May.

Markussen, T., K. van den Broeck, and F. Tarp (2011). ‘The Forgotten Property Rights:Evidence on Land Use Rights in Vietnam’. World Development, 39(5): 839–50.

Newman, C., K. van den Broeck, and F. Tarp (2014). ‘Social Capital, Network Effects,and Savings in Rural Vietnam’. Review of Income and Wealth, 60(1): 79–99.

Tarp, F. (1993). Stabilization and Structural Adjustment: Macroeconomic Frameworks forAnalysing the Crisis in sub-Saharan Africa. London and New York: Routledge.

Thurlow, J., F. Tarp, S. McCoy, N. M. Hai, C. Breisinger, and C. Arndt (2011). ‘TheImpact of the Global Commodity and Financial Crises on Poverty in Vietnam’.Journal of Globalization and Development, 2(1): Article 6.


25

2

Characteristics of the VARHS Dataand Other Data Sources

Kasper Brandt and Finn Tarp

2.1 Getting Familiar with VARHS

This chapter aims to illuminate the Viet Nam Access to Resources HouseholdSurvey (VARHS) database applied by the analytical studies in this book, cover-ing the period 2006 to 2014. We begin by reiterating the overall purposebehind VARHS, and then describe the data, including sample design, basiccharacteristics of the households interviewed, and attrition. Section 2.2, inturn, provides information on other databases and compares basic VARHSvariables to the 2009 population census and to the Viet Nam HouseholdLiving Standards Survey (VHLSS).

2.1.1 Purpose

The pilot of the first VARHS was designed in 2001–2. At the time, no existingsurvey provided—as detailed in Chapter 1—the background data needed forcoming better to grips with a series of intricate and pressing issues related tothe characteristics of existing rural markets for land, credit, and labour. More-over, there was limited information available on the way in which householdsaccessed resources in these markets. Little was known about the degree of (in)efficiency with which these markets operated, while the formulation ofevidence-based policy recommendations on their future development wasscant. This was so in spite of their critical role in the ongoing Doi Moi processof institutional and economic reform in a market-oriented direction.These kinds of concerns, combined with the illustrative information needs

already alluded to in Chapter 1, as well as the desire to monitor the impact of

ongoing policy and institutional changes referred to throughout this volume,inspired the formulation of the VARHS household questionnaire. It was expli-citly aimed at supplementing the information collected by General StatisticsOffice (GSO) in the nationally representative VHLSS, the successor of the VietNam Living Standard Survey (VLSS).1

The 2002 VARHS pilot was squarely focused on the rural areas of theprovinces of Ha Tay, Long An, Phu Tho, and Quang Nam. It covered some932 households, and it was clear from the beginning that available resourcesdid not permit VARHS to become a nationally representative survey. Instead,the purpose of VARHS was to start developing a unique panel of households.VHLSS has instead relied on a rotating panel of households; we reiterate fromChapter 1 that the VARHS and VHLSS are best understood as complementarysources of information. Each database has advantages and limitations.

2.1.2 Sample Design

The VHLSS 2002 was implemented throughout the year, and the VARHS2002 was designed to cover the exact same 960 households interviewedin the VHLSS in the first 6 months of 2002 in the 4 provinces listed inSection 2.1.1. Some 932 households were successfully identified and re-interviewed by Institute of Labour Science and Social Affairs (ILSSA) inNovember and December 2002. Except for attrition, the VARHS 2002 sampleis therefore representative for the rural areas of the four provinces.

Table 2.1 highlights the sample size in the different survey rounds ofVARHS. In 2006, the sample size increased to slightly over 2,300 rural house-holds living in 12 different provinces (Dak Lak, Dak Nong, Dien Bein, Ha Tay,Khanh Hoa, Lai Chau, Lam Dong, Lao Cai, Long An, Nghe An, Phu Tho, andQuang Nam)—to help readers visualize, see a general map of Viet Nam inFigure 2.1, Figure 2.2 showing the location of the 12 VARHS provinces, andFigure 2.3 showing the communes. Some 886 of the surveyed households in

Table 2.1 Sample size and survey months

2002 2006 2008 2010 2012 2014

Sample size New 932 1,438 1,011 – 553 4Total 932 2,324 3,277 3,208 3,704 3,648

Note: Last survey round’s ‘Total’ and current survey round’s ‘New’ do not add up to current survey round’s ‘Total’. This isdue to attrition households.

Source: VARHS data files.

1 See Glewwe, Agrawal, and Dollar (2004) for a series of studies based on the VLSS and Vietnamin the 1990s.

Characteristics of VARHS Data and Other Data Sources

27

Figure 2.1 Map of Viet NamSource: Recreated from Wikipedia (Wikimedia commons).


28

2006 were ‘survivors’ from the VARHS 2002, and 1,312 were householdssurveyed in the VHLSS 2004 Income and Expenditure sample.

Sample challenges emerged for three reasons: (i) GSO sampling was changedin 2004; (ii) several rural areas were reclassified as urban and administrativelysplit from 2004 to 2006; and (iii) standard attrition. For these reasons, 126randomly selected households were added to the VARHS, so the total numberin the VARHS 2006 amounted to 2,324 households.

It is important to emphasize that VARHS 2006 households were chosen toconstitute a representative sample of the rural areas in the twelve provinces inthat year. The subsequent survey rounds are not provincially representative

4 2

35

1

6

7

9

1011

8

12

4 2

35

1

6

7

9

1011

8

12

Provinces1 Ha Tay2 Lao Cai3 Dien Bieu4 Lai Chau5 Phu Tho6 Nghe Tho7 Quang Nam8 Khanh Hoa9 Dak Lak10 Dak Nong11 Lam Dong12 Long An

Figure 2.2 Location of the twelve VARHS provincesSource: VARHS data files.


29

in a strict statistical sense as they are based on households that were present inVARHS 2006. This includes that VARHS households were, in 2006, slightlyolder than a representative sample would have been.The 2008 and 2010 survey rounds were expanded in number and inter-

viewed more than 3,200 households. The approximately 1,000 new house-holds added in 2008 were included to evaluate the Danida-supportedAgricultural and Rural Development Sector Programme Support (ARDSPS)programme, located in the five provinces covered by the ARD-SPS programme(Lao Cai, Dien Bien, Lai Chau, Dak Lak, and Dak Nong).In addition to the over 3,200 households interviewed in 2010, the 2012

survey interviewed an extra 553 households, chosen with a view to ensuringbetter representativeness of the rural population in the surveyed provinces.

Figure 2.3 The VARHS communesSource: VARHS data files.


30

The reason for this extension of households is that the sample in VARHS, asnoted, was older than the representative sample in VHLSS because a largeshare of the VARHS households is limited to households that were presentin 2006.

To establish the balanced panel between 2006 and 2014, only householdsinterviewed in all survey rounds are included. This results in 2,162 house-holds, which can be followed over an eight-year time period. These householdsconstitute, with one exception, the base sample for the studies in this book.2

In some cases, however, the number of households may be smaller due tomissing data. The benefits from having panel data are substantial. Not onlycan aggregate changes over time be estimated more precisely than is possiblewith ‘repeated cross-sections’ (i.e. surveys of different households at differentpoints in time). One can also control for unobserved, time invariant house-hold characteristics in analytical work, and it is possible to investigateindividual-level changes over time. This means that the analyst can go beyondaggregate net changes in, say, landlessness, and ask specifically who gainedland, who lost land, and so on. This is critical in the present synthesis context,where a key aim is to uncover changes over time and understand their under-lying drivers.

In addition to the household survey, the VARHS also includes a commune-level survey. Interviews with commune administrators were performed in allcommunes where the VARHS households reside. This resulted in a commune-level panel database providing information on demographics, infrastructure,and local economic conditions. The commune panel includes 390 communesin the 12 provinces mentioned earlier in this section. These communes havebeen followed from 2006 to 2014.

2.1.3 Characteristics of the Balanced Sample

The balanced sample of 2,162 households is spread over 12 provinces and 464communes.3 Table 2.2 presents the location of the balanced sample house-holds. Almost 22 per cent of the sample live in the former province of Ha Tay(now part of Ha Noi), whereas less than 3 per cent live in LamDong. However,Lam Dong neighbours Dak Lak, Dak Nong, and Khanh Hoa, which are alsopart of the VARHS.

2 The one exception to the use of the balanced panel is Chapter 12 on children and youth. Inthis case, an unbalanced panel, also including 544 new younger households sampled from the2009 census to account for ageing of the original VARHS panel, is used. Adjusting the sample foryounger households that are more likely to have children was considered important in this caseto capture a complete picture of the evolution of children’s welfare over the sample period.

3 Although households are spread over 464 communes, the commune panel consists of 390communes as a consequence of commune attrition.


31

Furthermore, in some chapters the results are disaggregated by five regions.These regions are:

• Red River Delta: Includes VAHRS communes from the province of Ha Tay.In 2008, Ha Tay was subsumed into the metropolitan area of Hanoi. Theclose proximity to Hanoi means that urban-related activities such ashandicrafts contribute substantially to livelihoods. The location in theRed River Delta means that agriculture is focused on high-yield rice andvegetable production.

• North: Combines the Northeast and Northwest regions and includesVAHRS communes from the provinces of Lao Cai, Phu Tho, Lai Chau,and Dien Bien. These provinces are located in the more mountainous andremote areas of Northern Viet Nam on the borders to China and Laos.Except for Phu Tho, the provinces in the North are relatively poor. Theyalso exhibit low population densities of between 50–100 persons per km2,except Phu Tho where the population density is nearly 400 (GSO 2016).4

• Central Coast: Combines the North Central Coast and South Central Coastregions and includes VAHRS communes from the provinces of Nghe An,Quang Nam, and Khanh Hoa. This set of mountainous provinces on thecoast has a complex geography, including large areas covered by forest.They are dependent on agriculture, primarily rice, and a range of cashcrops such as rubber, cinnamon, peanuts, cashew, and coconuts. Inrecent years, some of these provinces have experienced high rates of

Table 2.2 Households and communes (balanced panel)

Province No. of households % of sample No. of communes % of sample

Ha Tay 470 21.7 68 14.7Lao Cai 85 3.9 24 5.2Phu Tho 297 13.7 44 9.5Lai Chau 109 5.0 29 6.3Dien Bien 99 4.6 28 6.0Nghe An 188 8.7 68 14.7Quang Nam 278 12.9 44 9.5Khanh Hoa 72 3.3 27 5.8Dak Lak 131 6.1 37 8.0Dak Nong 92 4.3 29 6.3Lam Dong 64 3.0 24 5.2Long An 277 12.8 42 9.1Total 2,162 100 464 100

Note: It is possible for households to move. This table is based on location in 2006.


4 Population density information is from 2014.


32

industrial and tourism growth. Population densities vary from between141 persons per km2 in Quang Nam to 229 in Khanh Hoa.

• Central Highlands: Includes VAHRS communes from the provinces ofDak Lak, Dak Nong, and Lam Dong. Placed on a series of contiguousplateaus surrounded by higher mountain ranges, households in thesecommunes are dependent on upland rice activities as well as a range ofcash crops, which are well suited to the higher altitudes and sub-tropicalclimate. Although there is a non-negligible production of products suchas tea, cocoa, and rubber, the main cash crop cultivated is coffee. Themajority of coffee produced in Viet Nam comes from this region (Luongand Tauer 2006). Population densities vary from 88 persons per km2 inDak Nong to 140 in Dak Lak.

• Mekong River Delta (MRD): Includes communes from the province ofLong An. Long An is located just west of the metropolitan area of Ho ChiMinh City. While not nearly as industrialized as the Southeast regionimmediately north of Ho Chi Minh City (not included in VARHS), theMRD has the third-highest industrial output of any region in Viet Namafter the Southeast region and the Red River Delta region. The MRD, alow-lying coastal region, is considered the rice bowl of Viet Nam. Eventhough the risk of flooding is severe, it has one of the highest outputs ofcereals per capita in Viet Nam. This also means the area supports a highpopulation density of 329 persons per km2.

We now focus attention on a few important variables and the geographicaldifferences between the twelve VARHS provinces. In focus here is the genderof the household head, age of the household head, ethnicity, education of allhousehold members aged 15 or above, and education of the household head.

Figure 2.4 illustrates the gender of household heads in the twelve provincesexamined in VARHS. The provinces where least households are headedby a male are Phu Tho (72.7 per cent), Quang Nam (68.3 per cent), KhanhHoa (66.7 per cent), and Long An (72.2 per cent). Five provinces have between75 per cent and 85 per cent of their households headed by amale. These are HaTay (75.1 per cent), Nghe An (77.1 per cent), and the three provinces in theCentral Highlands (82.4 per cent in Dak Lak, 78.3 per cent in Dak Nong,and 79.7 per cent in Lam Dong). The provinces in the North region (exceptfor Phu Tho) are the ones where most households are headed by a male. Here,85.9 per cent of households are headed by a male in Lao Cai, 86.9 per cent inDien Bien, and 91.7 per cent in Lai Chau.

Information on the age of household heads in 2014 is provided inFigure 2.5. The provinces in the Central Highlands and in the North (exceptfor Phu Tho) clearly have younger household heads compared to the rest ofthe examined provinces. The average age of household heads is between 50.5


33

and 51.5 years in all three provinces of the Central Highlands. In the North(except for Phu Tho), the average age of households varies between 49.2 yearsin Lai Chau to 53.7 years in Lao Cai (in Dien Bien the average age of householdheads is 52.1 years). The next interval, 55 to 58 years of age, is only representedby Ha Tay (56.4 years old) and Phu Tho (57.9 years old). The provinces withthe oldest household heads are Nghe An (58.4 years old), Long An (58.6 yearsold), Quang Nam (60.0 years old), and Khanh Hoa (60.7 years old).

City plots

65% – 75%75% – 85%85% – 92%

Male household head

City plots

3

4 2

51

6

7

9

1011

8

12

3

4 2

51

6

7

9

1011

8

12

Provinces1 Ha Tay2 Lao Cai3 Dien Bien4 Lai chau5 Phu Tho6 Nghe An7 Quang Nam8 Khanh Hoa9 Dak Lak10 Dak Nong11 Lam Dong12 Long An

Figure 2.4 Share of households with a male as household head in 2014Note: City plots are the 6 Vietnamese cities with more than 500,000 inhabitants according to the2009 population census (Ho ChiMinh City, Ha Noi, Da Nang, Hai Phong, Can Tho, and Bien Hoa).The thick black lines indicate region borders. It should be emphasized that Phu Tho (5) is includedin the North, Ha Tay (1) is part of the Red River Delta region, and the Central Coast region includesNghe An (6) in the North, Quang Nam (7) and Khanh Hoa (8) in the South.Source: VARHS data files.


34

Figure 2.6 shows the share of the balanced VARHS sample that is of Kinhethnicity disaggregated by the twelve provinces considered in VARHS. TheKinh people are the majority ethnic group in Viet Nam, andmade up approxi-mately 77 per cent of the population (IPUMPS 2016) in 2009. We clearly seethat the share of households that are Kinh varies between provinces. In theNorthern provinces (except for Phu Tho), the shares of households that are ofKinh ethnicity are as low as 10 per cent in Dien Bien, 14 per cent in Lai Chau,and 24 per cent in Lao Cai.

The reason for these low numbers of Kinh people in Dien Bien and Lai Chauis that another ethnic group constitute the majority—the Thai ethnic group.


City plotsCity plots

Age (households)

49.0 – 52.052.0 – 55.055.0 – 58.058.0 – 61.0

3

4 2

51

6

7

9

10

11

8

12

3

4 2

51

6

7

9

10

11

8

12

Figure 2.5 Average age of household head in 2014Note: See Figure 2.4, Note.Source: VARHS data files.


35

Lao Cai, on the other hand, is characterized by a large diversity of ethnicgroups, with the Kinh people among the best represented, together withHmong, Tay, and Dao. In the Central Highlands provinces, between 50 and75 per cent of households are Kinh. The rest of the provinces all have morethan 75 per cent of Kinh households, and three of the provinces almostexclusively consist of Kinh people (Ha Tay, Quang Nam, and Long An).Average years of schooling for all household members in rural areas aged

15 or above is illustrated in Figure 2.7, whereas average years of schooling for

City plots

0% – 25%50% – 75%75% – 100%

3

4 2

51

6

7

9

1011

8

12

3

4 2

51

6

7

9

1011

8

12

Kinh households

City plots


Figure 2.6 Percentage of Kinh households in balanced sample in 2014Note: See Figure 2.4, Note.Source: VARHS data files.


36

household heads is illustrated in Figure 2.8. Education for household membersaged 15 or above is, inwhat follows, referred to as the general level of education.5

Provinces in the North region (except for Phu Tho) have the least educatedrural areas, whereas Ha Tay and Phu Tho have the best educated rural zones.


City plots

5.0 – 6.06.0 – 7.07.0 – 8.08.0 – 9.09.0 – 10.0

3

4 2

51

6

7

9

1011

8

12

3

4 2

51

6

7

9

1011

8

12

Years of education (all)

City plots

Figure 2.7 Average years of schooling for all household members aged 15 or abovein 2014Note: The years of education is based on the answers to the question: ‘What grade did [NAME]finish?’ Each education category is associated with a number of years in school. City plots are the 6Vietnamese cities with more than 500,000 inhabitants according to the 2009 population census(Ho Chi Minh City, Ha Noi, Da Nang, Hai Phong, Can Tho, and Bien Hoa). The thick black linesindicate region borders. It should be emphasized that Phu Tho (5) is included in the North, Ha Tay(1) is part of the Red River Delta region, and the Central Coast region includes Nghe An (6) in theNorth, Quang Nam (7) and Khanh Hoa (8) in the South.Source: VARHS data files.

5 It is worth reiterating that the focus is here on rural areas, ignoring any influence caused byurban people.


37

The gap in general level of education is more than four years between thelowest educated province (Lai Chau) and the two highest educated provinces(Ha Tay and Phu Tho). The gap is even higher when only considering house-hold heads. The gap between the lowest educated and the highest educatedprovince is more than five years.Only Lao Cai, with an average of 6.2 years, has a general level of education

between 6 and 7 years. This indicates that the Northern provinces (except forPhu Tho) are substantially less educated compared to the rest of the examined


City plotsCity plots

ears of education (head)Y

3

4 2

51

10

11

8

12

3.8 – 5.06.0 – 7.07.0 – 8.08.0 – 9.0

6

7

9

3

4 2

51

10

11

8

12

6

7

9

333333333333333

4444

55551111111111111

0001111

111111

8888888

11111111

66666

77

99999

Provinces1 Ha Tay2 Lao Cai3 D e en

c au5 Phu Tho6 Ng An7 Quang am8 Khanh Hoa9 ak Lak10 Dak Nong11 Lam Dong12 Long n

City plotCity plot

3.8 – 5.06.0 – 7.07.0 – 8.08.0 – 9.0

ProvincesProvincesProvinces11 Ha1 Ha TaTayy22 Lao Ca2 Lao Caiii3 Dien Bien3 Dien Bien3 Dien Bien4 L h4 Lai chau4 Lai chau5 Phu h5 Phu Th5 Phu Thoo6 N h An6 Nghe An6 Nghe An7 Quan a7 Quang Na7 Quang Namm8 Khanh Hoa8 Khanh Hoa8 Khanh Hoa9 D k9 Dak La9 Dak Lakkk10 Dak Nong10 Dak Nong10 Dak Nong11 Lam Dong11 Lam Dong11 Lam Dong12 Long An12 Long An12 Long An

City plotsCity plotsCity plotsCity plotsCity plotsCity plots

YY d)d)ears of education (headears of education (headee

3333

2444

555111

666

777

9999

00011

1111

88888

2111

3333333333333333333333333333333333

4 24 2222244 2244444 2222222222244444

5555555555555555511111111111111

6666666666666666666666

77777777

99999999999999999999999

00000000011100000001110000000000111111

11111111111111111

888888888888888888888888888888

22222222222222222222222222222221112212112222222221111

3 8 – 5.03.8 – 5.03.8 – 5.06.0 – 7.06.0 – 7.06.0 – 7.07.0 – 8.07.0 – 8.07.0 – 8.08.0 – 9.08.0 – 9.08.0 – 9.0

Figure 2.8 Average years of schooling for household heads in 2014Note: The years of education is based on the answers to the question: ‘What grade did [NAME]finish?’ Each education category is associated with a number of years in school. City plots are the 6Vietnamese cities with more than 500,000 inhabitants according to the 2009 population census(Ho Chi Minh City, Ha Noi, Da Nang, Hai Phong, Can Tho, and Bien Hoa). The thick black linesindicate region borders. It should be emphasized that Phu Tho (5) is included in the North, Ha Tay(1) is part of the Red River Delta region, and the Central Coast region includes Nghe An (6) in theNorth, Quang Nam (7) and Khanh Hoa (8) in the South.Source: VARHS data files.


38

provinces. In the Central Highlands, two out of three provinces have a generallevel of education between seven and eight years, although these are veryclose to eight years. Also, KhanhHoa in the Central Coast has a general level ofeducation between seven and eight years. Only Long An in the MRD regionhas a general level of education between eight and nine years. The fiveremaining provinces (Dak Nong, Quang Nam, Nghe An, Phu Tho, and HaTay) all have a general level of education between nine and ten years.

Years of education for rural household heads are slightly lower than thegeneral level of education. The explanation is that household heads tend to beolder than the average population and younger people are generally bettereducated. The provinces with the lowest average years of schooling for house-hold heads are Lao Cai (3.8 years), Lai Chau (3.9 years), and Dien Bien(4.7 years), even though these provinces, on average, have relatively younghousehold heads. While no province has an average years of schoolingfor household heads between five and six years, three provinces have anaverage between six and seven years. These are Long An (6.1 years), Lam Dong(6.6 years), and Khanh Hoa (6.3 years).

Next, two provinces have an average years of schooling for household headsbetween seven and eight years. Quang Nam has an average of 7.1 years,whereas Dak Lak has an average of 7.4 years. The provinces with the highesteducated household heads are Ha Tay (8.7 years), Dak Nong (8.7 years), NgheAn (8.8 years), and Phu Tho (9 years).

2.1.4 Sample Attrition

When households drop out and are not re-interviewed in later survey rounds,we say that the survey is subject to sample attrition.6 In household surveys likeVARHS it is inevitable to have sample attrition as some households refuse to bere-interviewed or when all members of a household die. Both of these reasonsfor attrition are rare in VARHS. Amore common reason for attrition ismigration.Based on responses from local authorities, two thirds of migrating householdsare believed to have migrated permanently, whereas one third is understood tohave migrated temporarily. When households migrate it may be too costly oreven impossible to relocate them. Whether migrating households differ signifi-cantly from remaining households is of importance, as leaving out a specificpopulation group from the sample may bias analytical results.

As presented in Table 2.1, 2,324 households were interviewed in 2006.In 2014, 2,162 of the very same households interviewed in 2006 had beeninterviewed in all subsequent survey rounds. The attrition rate in each survey

6 Chapter 10 includes further analysis of attrition.


39

round is provided in Table 2.3. The attrition rates between one wave andthe next vary between 1.1 per cent and 2.5 per cent. The overall attritionrate from 2006 to 2014 is 7 per cent. This is fairly low considering that thereare five waves.Table 2.3 further investigates whether household heads in attrition house-

holds are older or younger than household heads of those remaining in thesurvey. In addition, Table 2.3 compares average years of education of house-holdmembers aged 15 and above and the share of households that are of Kinhethnicity for attrition and non-attrition households. Household heads inattrition households are, on average, older than the remaining householdheads. For the last survey round, however, the difference is only borderlinesignificant on a 10 per cent significance level.Differences in households’ average years of education are only significant

for two survey rounds. The households that were not re-interviewed in 2008and the ones that were not re-interviewed in 2012 are, on average, worseeducated than the remaining households. From Table 2.3, attrition does notseem to correlate with ethnicity, as no systematic differences occur. The lowrate of attrition is, to a large extent, a testament to the professionalism of theVARHS implementation and to the low numbers of migrating households inrural Viet Nam.

Table 2.3 Extent of attrition and comparison of attrition and non-attrition households

Samplesize

Numberattrited

Mean:attrited

Mean:non-attrited

t test forNA-A=0

Age of household head2006 base 2,3242006-8 panel 2,266 58 46.2 39.8 4.062006-8-10 panel 2,225 41 45.7 39.7 3.262006-8-10-12 panel 2,187 38 43.3 39.6 1.962006-8-10-12-14 panel 2,162 25 43.4 39.6 1.69

Average years of education2006 base 2,3242006-8 panel 2,266 58 4.8 6.1 �2.632006-8-10 panel 2,225 41 5.6 6.1 �0.922006-8-10-12 panel 2,187 38 4.8 6.2 �2.322006-8-10-12-14 panel 2,162 25 6.3 6.2 0.23

Kinh households (%)2006 base 2,3222006-8 panel 2,264 58 79.3 80.4 �0.212006-8-10 panel 2,223 41 80.5 80.4 0.012006-8-10-12 panel 2,185 38 73.7 80.5 1.062006-8-10-12-14 panel 2,160 25 80.0 80.6 �0.07

Note: The average years of education is only for household members aged 15 or above. Two households did not provideinformation on ethnicity, and hence, the sample size is 2,322 in 2006 instead of 2,324.



40

Based on responses from local authorities, the VARHS provides informationon the perception of why households are absent and if these households wererelatively poor or rich within the community. The local authorities werefurther asked about the occupation and standard of living for the absenthouseholds. The most common reasons for migrating are economic or toreunite with family members. Further, the destination is typically anotherprovince or a non-bordering district, and more households are believed tohave moved to another rural area than to an urban area.

As noted, Chapter 10 takes the attrition analysis further in order to revealwhether the standard of living differs between the attrition and non-attritionhouseholds.

2.2 VARHS and Other Data Sources

The studies in this book are primarily based on VARHS. Other data sources do,however, exist. It is therefore of interest to enquire whether VARHS and theother data sources are in accordance with each other. We begin by noting thatVARHS was initially a sub-sample of the households which participated in the2002 VHLSS, chosen to be provincially representative.

2.2.1 Other Data Sources

POPULATION CENSUSESPopulation censuses were conducted in Viet Nam in 1999 and 2009 (andbefore that in 1979 and 1989). The intention of the most recent census in2009 was to gather information on the population and housing situation forthe entire territory of Viet Nam—both nationally and locally (GSO 2010a).The complete population census consisted of basic questions about the indi-vidual, education, ethnicity, and housing situation. In addition to the com-plete population census, a sample of 15 per cent was asked further questionsregarding prior place of residence, marriage, fertility, death, and householdassets. This 15 per cent sample constitutes one of the two sources of informa-tion to which VARHS is explicitly compared in this chapter.

The whole country was for the 2009 census divided into 172,000 enumer-ation areas with an average size of 100 households. Around 60,000 trainedsurveyors took part in the survey which lasted from 1 to 20 April. Due to thescale of the survey, the population census serves as the most reliable sourceof basic information about the population. However, the information isnot very detailed and more in-depth surveys are needed for specific studiesof interest.


41

VIET NAM HOUSEHOLD LIVING STANDARDS SURVEY (VHLSS)Since 1993, the VHLSS (and its predecessor VLSS) has collected information onliving standards in Viet Nam. From 2002 onwards, the survey has been con-ducted biennially (i.e. every two years). For every survey round the overallsurvey is divided into three sub-surveys: (i) a core questionnaire to the major-ity of households (80 per cent of the sample size goes into this group); (ii) corequestions and more detailed questions on specific topics to a minority ofhouseholds (20 per cent of the sample size goes into this group); and (iii) acommunity survey.The sample design of VHLSS relies on a master sample based on the most

recent population census, and it is stratified by province and urban/ruralstatus. Until 2008, the master sample consisted of 3,063 communes, andfrom 2010 the master sample increased the number of communes to 3,133.The next step in selecting the sample is to select three enumeration areas (EAs)per commune. Both communes and EAs are selected based on their size(number of households according to most recent population census), mean-ing that communes and EAs with a relatively large number of households havea higher probability of being selected for the sample.Next, households are randomly sampled from the master sample and

weights on households are calculated to represent the most recent populationcensus for some basic characteristics. The sample is designed as a rotatingpanel where half of the EAs are replaced each survey round and new EAsfrom the same communes enter the survey. These new EAs must not havebeen selected in the two most recent survey rounds.The rotating design is summarized in Figure 2.9. In half of the retained

communes, households were interviewed in the two previous rounds, whereashouseholds were interviewed only in the previous survey round in the otherhalf of the retained communes. This implies that, in each survey round, halfare new households, whereas one fourth was also present in the last surveyround, and one fourth present in the last two survey rounds.For the survey rounds based on the 1999 population census, 3,063 com-

munes were chosen out of a total of 10,476 communes (GSO 2004, 2006,2008). The total sample size was slightly less than 46,000 households, whichcorresponds to 15 households per commune. Of these, around 36,700 werechosen for the first group of households receiving only the core questionnaire,and slightly less than 9,200 households were chosen for the second group,receiving the core questionnaire and an extended survey on specific topics.7

In 2010 and 2012, the number of communes in the survey increased to3,133 (GSO 2010b). The sample size increased significantly in 2010 to 69,360

7 In 2002, however, the sample size included 75,000 household from which 45,000 received thecore questionnaire and 30,000 also received the extended expenditure survey.


42

households, because a new master sample was created based on the 2009population census, while at the same time re-interviewing half of the samplehouseholds in 2008. In 2012 the sample size, naturally, fell to 46,995 house-holds, which corresponds to 15 households per commune.

A key objective of the VHLSS is to measure consumption poverty rates andother living standard indicators, whereas VARHS is focused on providing dataon land, credit, and labour over time. In addition to poverty measures, theVHLSS provides information on health, education, employment, householdassets, infrastructure, and general information on the commune. Anotherstated objective of the VHLSS has been to serve as a monitoring tool for specificprogrammes and contribute to evaluating the realization of the MillenniumDevelopment Goals (MDG).

ADDITIONAL DATA SOURCESOther data sources exist as well. They include the Labour Force Survey (LFS),Community-Based Monitoring System (CBMS), the Informal Sector Survey(ISS), Young Lives, Demographic and Health Surveys (DHS), World ValuesSurvey (WVS), and a couple of different informal enterprise surveys. Anotherhousehold panel dataset covering the provinces of Dak Lak, Ha Tinh, andThua Thien Hue was also initiated over the period 2007–10 by Phung DucTung as part of his PhD research (Thung 2012).

Every year since 2009, the LFS has been conducted based on the 15 per centsample of the 2009 population census. The purpose of this survey is to collectinformation on the labour market in Viet Nam. Specifically, information onthe total size and distribution of the labour force, the unemployment andunderemployment for different population groups, composition of labourforce by various categories like occupation and industry, working conditions,economically inactive people, and migrants are of interest.

The CBMS was carried out in Viet Nam in 2006 covering fifty-two com-munes in five provinces (Ha Tay, Yen Bai, Quang Ngai, Lam Dong, and Ninh

Round 1

Group 1

Group 2

Group 3

Group 4

Round 2

Group 2

Group 3

Group 5

Group 6

Round 3

Group 3

Group 5

Group 7

Group 8

Round 4

Group 5

Group 7

Group 9

Group 10

Figure 2.9 Rotating sampling design of VHLSSSource: Authors’ illustration.


43

Binh). The main objective of the survey was to improve the monitoring ofpoverty in Viet Nam, and to enhance the capacity of official planners toalleviate poverty. In total, 57,884 households were surveyed and asked generalquestions about their characteristics, land and asset ownership, education,health, and income of the household.The Young Lives Project was designed to follow 12,000 children in four

different countries, of whom 3,000 are from Viet Nam (Young Lives 2014).The children were initially surveyed in 2002, and they are to be followed forfifteen years, during which time they will be surveyed five times. The mostrecent survey round (the fourth) took place in 2013. The sites examined arelocated in Lao Cai, Hung Yen, Phu Yen, Ben Tre, and in areas around the cityof Da Nang.The purpose of the WVS has been to help understand better changing

beliefs, equality, subjective wellbeing, economic development, and othersocioeconomic issues (WVS 2016). The WVS was last implemented in VietNam in 2006, and the country did not participate in the most recent surveyround. In 2006, however, 1,495 individuals throughout the country aged 18and above participated in the survey.The DHS Programme has been implemented three times in Viet Nam—in

1997, 2002, and 2005. The surveys deviate from each other by containingdifferent health modules like alcohol consumption, iodine salt-testing, cook-ing fuel, and information on malaria. The two first surveys provided informa-tion on HIV associated with behaviour and knowledge, whereas the mostrecent survey extended the HIV module to include HIV testing. Yet furthersurveys exist on enterprise related issues. They include the Small andMedium-Scaled Enterprise (SME) Survey, the Provincial Competitiveness Index (PCI),the Viet Nam Enterprise Survey (VES), and the Viet Nam Enterprise Census(VEC).8

All of the data sources mentioned in this section provide specific data thatmay be of use in analytical work. However, the only two sources that can, in ameaningful way, be compared to VARHS, are the VHLSS and the populationcensus. We now turn to this comparison.

2.2.2 Comparison of VARHS with VHLSS and the Population Census

This sub-section compares characteristics of household heads in the 2006VARHS with household heads in the 15 per cent sample of the 2009 popula-tion census (referred to as the 2009 population census in what follows) and inVHLSS 2006.

8 See <http://eng.pcivietnam.org/> and <https://www.gso.gov.vn/Default_en.aspx?tabid=491>.


44

http://eng.pcivietnam.org/

https://www.gso.gov.vn/Default_en.aspx?tabid=491

Table 2.4 presents information on gender, age, illiteracy, and ethnicity ofhousehold heads in VARHS 2006 compared to VHLSS 2006 and the 2009population census. Small differences exist for the gender variable. ComparingVARHS with the 2009 population census, only Khanh Hoa has a deviationof more than 5 percentage points (5.2 percentage points). In addition, there isno systematic bias between VARHS and the two other data sources—for sevenout of twelve provinces, the share of male-headed households is larger inVARHS compared to VHLSS and the population census.

The next variable, age of household head, appears more biased. For everyprovince, except for Lam Dong, the average age is highest in VARHS andsecond-highest in VHLSS.9 This age difference is, however, as expectedbecause VARHS households are ‘survivors’ from the VARHS 2002 and a sampleof households from the VHLSS 2004. Consequently, newly established andyounger households are not present in VARHS 2006. In Figure 2.5 we foundthat the provinces in the North (except for Phu Tho) and the provinces inthe Central Highlands had considerably younger household heads in 2014compared to the rest of the provinces. This characteristic is clearly alsopresent in VARHS 2006, which can be verified by the two other data sources(cf. Table 2.4).

In the VARHS 2006, around 34.6 per cent of household heads are below 40years of age in the ‘young’ provinces (provinces in the North, except for PhuTho, and in the Central Highlands), whereas the same figure is a mere 18.2 percent in the ‘old’ provinces (Ha Tay, Phu Tho, Nghe An, Quang Nam, KhanhHoa, and Long An). At the other end of the scale, 14.8 per cent of householdheads are aged 60 or above in the ‘young’ provinces, whereas the correspond-ing number for the remaining provinces is 28.4 per cent. Similarly, in VHLSS2006 and the 2009 population census, provinces in the North (except for PhuTho) and provinces in the Central Highlands have a much larger fraction oftheir household heads being less than 40 years of age and a lower fraction aged60 or above. As is also evident in VARHS 2006, the fraction of household headsin the age group between 40 and 59 is fairly similar across ‘young’ and ‘old’provinces in both VHLSS 2006 and the 2009 population census.

As illustrated in Figure 2.6, there are substantial differences in ethnicityamong provinces based on VARHS information. Table 2.4 shows that thesedifferences are present in other data sources as well. The share of householdheads that are of Kinh ethnicity deviates by no more than 5 percentage pointsfrom VARHS to at least one of the two other data sources (except for LamDong, where it deviates by 7.5 percentage points to the census). It should,however, be mentioned that, for Nghe An and Quang Nam, the share of

9 Lam Dong is the province with the fewest households (cf. Table 2.2). Thus, the scope foruncertainty is high.


45

household heads that are of Kinh ethnicity is substantially larger in VARHS2006 compared to the census in 2009. In the VARHS 2006 the shares are 17.2and 19.2 percentage points higher in Nghe An and Quang Nam, respectively,compared to the 2009 population census.The last variable of interest in Table 2.4 is illiteracy. Overall, the VARHS

figures resemble the VHLSS and census figures. Household heads are consid-erably more likely to be illiterate in the North (except for Phu Tho), where theilliteracy rate reaches as much as 46 per cent in VARHS and VHLSS. Also,within the North region (except Phu Tho), the data sources are ordering theprovinces similarly as Lai Chau has the highest illiteracy rate followed by DienBien and at last Lao Cai. Next, the illiteracy rate in the Central Highlandsand Khanh Hoa is still more than 10 per cent in the 2009 population census.

Table 2.4 Comparison of gender, age, ethnicity, and illiteracy for household (hh) heads

Male (hh heads) Age (hh heads)

2006 VARHS(%)

2006 VHLSS(%)

2009 census(%)

2006 VARHS 2006 VHLSS 2009 census

Ha Tay 78.6 78.3 77.1 50.7 49.6 46.3Lao Cai 88.9 87.9 88.3 47.5 45.1 39.9Dien Bien 90.2 90.5 88.5 46.5 42.3 39.9Lai Chau 92.2 91.1 90.6 43.6 42.1 39.0Phu Tho 78.5 78.3 77.8 52.1 49.6 47.2Nghe An 83.2 81.9 81.9 52.2 51.3 46.8Quang Nam 72.7 77.8 74.2 54.4 51.4 49.3Khanh Hoa 68.8 67.9 74.0 54.9 47.9 46.4Dak Lak 82.5 88.3 84.8 46.9 44.2 43.2Dak Nong 84.3 86.9 87.7 45.7 42.4 40.6Lam Dong 79.4 80.8 81.1 44.9 47.0 43.5Long An 74.6 71.4 72.2 52.3 50.8 47.7

Kinh ethnicity (hh heads) Illiteracy (hh heads)

2006 VARHS(%)

2006 VHLSS(%)

2009 census(%)

2006 VARHS(%)

2006 VHLSS(%)

2009 census(%)

Ha Tay 99.0 99.5 98.5 5.5 5.1 3.1Lao Cai 23.3 30.3 24.4 20.0 16.7 28.6Dien Bien 7.1 10.7 10.8 36.6 45.2 34.0Lai Chau 13.9 3.3 11.2 46.1 46.7 39.9Phu Tho 82.2 82.2 81.5 1.9 5.4 2.7Nghe An 89.8 89.2 72.6 6.6 6.4 6.6Quang Nam 98.7 94.8 79.4 7.4 10.4 9.6Khanh Hoa 92.2 92.6 81.9 10.4 12.3 13.7Dak Lak 71.3 75.7 64.9 14.0 12.6 10.5Dak Nong 76.9 78.6 66.2 5.6 3.6 10.3Lam Dong 63.2 73.1 70.7 19.1 11.5 11.1Long An 99.7 100.0 99.8 7.0 10.3 7.3

Note: For the 2009 population census and VHLSS 2006, only rural households are included. The figures of illiteracy are forpeople above 5 years old, and it is assumed that people who have finished primary school are literate.

Source: VARHS data files, VHLSS data files, and a 15 per cent sample of the population census 2009.


46

As an outlier, however, Dak Nong has a relatively low illiteracy rate in theVARHS and VHLSS data. Except for Dak Nong, the three different data sourcesare consistent on the provinces with the lowest illiteracy rates (Phu Tho, HaTay, and Nghe An).

2.3 Conclusion

The purpose of this chapter was to start familiarizing the reader with theVARHS database, present geographical variation among provinces for keyvariables, introduce other data sources, and compare VARHS to two provin-cially representative databases.

Overall, the VARHS database has been very successful in tracking the initialhouseholds from 2006, leaving analysists with a unique panel dataset. Illus-trating key variables in map format, we saw that large variation occurs amongprovinces. Especially, the provinces in the North region (except for Phu Tho)are more male dominated, the household heads are relatively young, fewerhouseholds are of Kinh ethnicity, and they have much less education com-pared to the rest of the examined provinces.

Finally, we saw in this chapter that VARHS households are—except asexpected for age—quite similar to households in the VHLSS and the popula-tion census. This suggests that the VARHS and the results put together in thisbook are likely to apply to amore extensive part of the Vietnamese populationthan initially planned for.

References

Glewwe, P., N. Agrawal, and D. Dollar (eds) (2004). Economic Growth, Poverty, andHousehold Welfare in Vietnam. World Bank Regional and Sector Studies 29086. Avail-able at: <https://openknowledge.worldbank.org/bitstream/handle/10986/15010/290860rev.pdf?sequence=1>. Accessed 8 May 2016.

GSO (General Statistics Office of Vietnam) (2004). Vietnam: Household Living Stand-ards Survey 2004. Available at: <http://microdata.worldbank.org/index.php/catalog/2370>. Accessed 8 May 2016.

GSO (General Statistics Office of Vietnam) (2006). Result of the Survey on HouseholdLiving Standards 2006. Available at: <http://www.ilo.org/surveydata/index.php/catalog/86>. Accessed 8 May 2016.

GSO (General Statistics Office of Vietnam) (2008). Result of the Survey on HouseholdLiving Standards 2008. Available at: <http://www.ilo.org/surveydata/index.php/catalog/77>. Accessed 8 May 2016.


47

https://openknowledge.worldbank.org/bitstream/handle/10986/15010/290860rev.pdf?sequence=1

https://openknowledge.worldbank.org/bitstream/handle/10986/15010/290860rev.pdf?sequence=1

http://microdata.worldbank.org/index.php/catalog/2370

http://microdata.worldbank.org/index.php/catalog/2370

http://www.ilo.org/surveydata/index.php/catalog/86




GSO (General Statistics Office of Vietnam) (2010a). The 2009 Vietnam Population andHousing Census: Major Findings. Available at: <https://www.gso.gov.vn/default_en.aspx?tabid=515&idmid=5&ItemID=9813. Accessed 8 May 2016.

GSO (General Statistics Office of Vietnam) (2010b). Result of the Vietnam HouseholdLiving Standards Survey 2010. Available at: <http://www.gso.gov.vn/default_en.aspx?tabid=515&idmid=5&ItemID=12426>. Accessed 8 May 2016.

GSO (General Statistics Office of Vietnam) (2016). Population and Employment. Avail-able at: <http://www.gso.gov.vn/default_en.aspx?tabid=774>. Accessed 8 May 2016.

IPUMPS (Integrated Public Use Microdata Series) (2016). Data available at: <https://international.ipums.org/international-action/variables/group>. Accessed 8 May 2016.

Luong, Q. V. and L. W. Tauer (2006). ‘A Real Options Analysis of Coffee Planting inVietnam’. Agricultural Economics, 35(1): 49–57.

Thung, P. D. (2012). Vulnerability to Poverty in Rural Vietnam. Saarbrücken: LAMBERTAcademic Publishing.

Young Lives (2014). Young Lives Survey Design and Sampling in Vietnam. Available at:<http://www.younglives.org.uk/content/young-lives-survey-design-and-sampling-vietnam>. Accessed 8 May 2016.

WVS (World Values Survey) (2016). Information available at: <http://www.worldvaluessurvey.org/WVSContents.jsp>. Accessed 8 May 2016.


48

https://www.gso.gov.vn/default_en.aspx?tabid=515&e_x0026;idmid=5&e_x0026;ItemID=9813

https://www.gso.gov.vn/default_en.aspx?tabid=515&e_x0026;idmid=5&e_x0026;ItemID=9813

http://www.gso.gov.vn/default_en.aspx?tabid=515&e_x0026;idmid=5&e_x0026;ItemID=12426


http://www.gso.gov.vn/default_en.aspx?tabid=774

https://international.ipums.org/international-action/variables/group

https://international.ipums.org/international-action/variables/group

http://www.younglives.org.uk/content/young-lives-survey-design-and-sampling-vietnam

http://www.younglives.org.uk/content/young-lives-survey-design-and-sampling-vietnam

http://www.worldvaluessurvey.org/WVSContents.jsp

http://www.worldvaluessurvey.org/WVSContents.jsp

Part IA Rural Economy Transformation

3

Local Transformation

A Commune-Level Analysis

Ulrik Beck

3.1 Introduction

The process of structural transformation takes place at many levels. At one endof the spectrum, it is the result of decision-making of individual households oreven household members. At the other end of the spectrum, governmentpolicies can affect the direction and speed of transformation.

There are, however, several intermediate levels, which form part of theframework within which households act. This is a part of the transformationprocess that runs the risk of being overlooked if one relies solely on datacollected at the more disaggregated household level. The commune, thelowest administrative division in Viet Nam, is a natural level of analysis forproviding a high-level yet local view of changing economic conditions and ofthe structural transformation that has taken place in the last eight years.Vietnamese communes typically consist of a few separate villages and, in2014, the average number of households in communes participating in theViet Nam Access to Resources Household Survey (VARHS) was 2,079. This size,combined with the fact that long-distance travel in rural Viet Nam stillrequires a significant commitment of both time and money, means that theconditions of the commune of residence are informative about the everydayconditions faced by rural Vietnamese households.

In order to clarify the framework conditions under which households oper-ate, and in order to illuminate the transformation process at the local level, theVARHS includes a commune questionnaire, which has been administered toadministrators in all communes in which the VARHS is collected. This study

utilizes the resulting commune-level panel database to provide an overview ofeconomic conditions and transformation in the years 2006–14. The finalsection of the chapter looks ahead by pointing to some potential futurechallenges for the VARHS communes and for the people living in them.The commune panel includes 390 communes that were interviewed every

second year from 2006 to 2014. Table 3.1 shows how the communes aredistributed within the five regions. It also shows how communes are distrib-uted within three income tertiles. It is immediately clear that there aredifferences both within and between the five regions. In the provinces ofthe Red River Delta and the Mekong River Delta (MRD), which are close tothe large population centres of Hanoi and Ho Chi Minh City, many communesare doing quite well. This is especially the case for the only province in thesample that belongs to the MRD, namely the province of Long An. Here,more than two-thirds of communes belong to the highest (third) incometertile and this region has the highest average income per capita of the fiveregions. Conversely, in the predominantly remote and mountainous Northregion, more than two-thirds of communes belong in the lowest (first)income tertile. Within the North region, the province of Phu Tho is doingmuch better than the remaining, more remotely located Northern provinces.If Phu Tho is removed from the North region, 85 per cent of the remainingNorth communes fall into the lowest income tertile. The Central Coastregion is doing markedly better than the North region, but not quite aswell as the Central Highlands region where most communes are in thehighest income tertile. The communes of the Central Highlands region are,on average, doing better than the Red River Delta communes in terms of per

Table 3.1 Communes in commune panel sample by 2014 income tertile and region

Red RiverDelta

North CentralCoast

CentralHighlands

MekongRiver Delta

Total

1 12 69 37 12 5 135(18.5) (69.) (33.9) (15.6) (12.8) (34.6)

2 29 21 48 20 7 125(44.6) (21.0) (44.0) (26.0) (18.0) (32.1)

3 24 10 24 45 27 130(36.9) (10.0) (22.0) (58.4) (69.2) (33.3)

Total 65 100 109 77 39 390Average monthly incomeper capita in 2014,commune average=100

101.5 74.5 96.8 121.2 129.6 100.0

Notes: Income tertiles are based on stated average commune income. Due to bunching of answers, there are not exactly⅓ of communes in each of the three tertiles. Column frequencies in per cent are reported in parentheses. Income iscalculated as an unweighted mean of the average per capita income in each commune.

Source: Author’s calculations based on VARHS database.

52

A Rural Economy Transformation

capita income. It should be kept in mind that income is not necessarilyequivalent to consumption if there is substantial consumption of own pro-duction. The high prevalence of cash crop agriculture in the Central High-lands decreases that wedge between income and consumption. This can helpto explain why the Central Highlands appear to be doing so well comparedto the other regions in this table. Even though this is not the case for allindicators of welfare, in 2010 only the Northern Mountain regions and theNorth Central Coast had higher poverty rates than the Central Highlandsregion (World Bank 2012).

Figure 3.1documents the evolutionof thenumberofhouseholds in theaveragecommune in the sample. Communes in the North tend to be smaller thanelsewhere, but in all regions communes have been growing in terms of numberof households over the period. This reflects the general population increase inViet Nam over the period. Even though there has been a tendency of migrationtowards the urban areas in the last decade, an increase in the number of ruralhouseholds is still apparent in the rural communes of the VARHS sample.

3.2 Occupational and Agricultural Choice

This section aims to provide an overview of what the households in theVARHS communes are doing for a living. Figure 3.2 shows the evolution ofthe most important occupations in the communes. In almost all of them,agriculture is one of the three most important occupations, with upwardsof 90 per cent of all households taking part. Aquaculture, other services,construction activity, and other occupations have all gained importance

0

500

1000

1500

2000

2500

3000

2006 2008 2010 2012 2014

Ave

rag

e n

umb

er o

f H

Hs

Year

Red River Delta North Central Coast Central Highlands Mekong River Delta

Figure 3.1 Average number of households in VARHS communes by regionSource: Author’s calculations based on VARHS database.


53

over the period.1 The increase in aquaculture as an important occupationcorresponds well to a well-documented increase in aquaculture productionin Viet Nam (see, for instance, the Food and Agricultural Organization (FAO)Fisheries and Aquaculture Statistics Database). The increase in constructionactivity reflects the high levels of economic growth experienced in Viet Namover the period. While the occupational shifts mentioned are statisticallysignificant and point towards structural transformation, the occupation struc-ture has not changed radically. The picture that emerges is instead one ofdiversification at the commune level into a wider range of activities withoutleaving the main occupation of agriculture behind.The country-level averages can hide interesting geographical variation.

In order to explore this, Figure 3.3 shows the most important occupations in2014 by region.While agriculture is important in all regions, it is slightlymoreimportant in the more remote and poorer Northern provinces as well as theCentral Highlands. In these two regions, almost 100 per cent of communesreport that agriculture is one of the most important occupations. In the moresparsely populated Northern region, more than 50 per cent of communesengage in forestry while almost no communes in the more densely populated

0

10

20

30

40

50

60

70

80

90

100

2006 2008 2010 2012 2014

Perc

enta

ge

of

com

mun

es

Year

Agriculture Forestry Aquaculture HandicraftsConstruction Other services Other occupations

Figure 3.2 Most important occupations by year, % of communesNote: The graph shows the share in % of communes where different occupations were among thethree most important occupations. Commune officials were asked to mention the three mostimportant occupations. Officials had the option to list fewer than three if there were not threerelevant occupations. ‘Other occupations’ include everything not included in the other categories,including transport and manufacturing.Source: Author’s calculations based on VARHS database.

1 The changes in occupational choice are all statistically significant at the 5 per cent level, using atwo-sided t-test comparing 2006 and 2014 occupations at the commune level.


54

provinces in the Red River Delta and MRD regions do so. Handicrafts andother occupations are more common occupations around the big populationcentres as well, and in the specific provinces of VARHS, handicrafts is a par-ticularly common occupation in the Red River Delta area of ex-Ha Tay. Manycommunes in the MRD province of Long An engage in activities that fallunder the category of other occupations, including transport and manufac-turing, both of which are typical of rural areas in close proximity to largeurban population centres.

Since agriculture is the most important occupation throughout the periodand in all regions, it is worth digging deeper into the structure of agriculture.Figure 3.4 shows how the allocation of land for different uses varies betweenregions as well as over time. In the Red River Delta region, the majority of landis used for rice cultivation. This share has steadily declined over time, how-ever. Instead, more and more land is used for non-rice annuals as well as forresidential purposes.

In the North, there has been a steady decline in forested land while the sharesof all other land uses have increased. Deforestation has also taken place in theCentral Coast region. This land hasmostly been converted into residential land.The North and Central Coast regions had high initial forestation rates (morethan 50 per cent and 30 per cent in 2006, respectively). As population densities

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

Agriculture Forestry Aquaculture Handi-crafts

Construction Trade Otherservices

Otheroccupations

Perc

enta

ge

of

com

mun

es

Sector


Figure 3.3 Most important occupations by region in 2014, % of communesNote: See Figure 3.2, Note.Source: Author’s calculations based on VARHS database.


55

and income levels rise, some of this land has been converted into agriculturaland residential land. In the North, which is still heavily focused on agriculture,as shown in Figure 3.3, much of the newly cleared land was converted intoagricultural uses. As shown, construction and other activities such as tourismhave become increasingly important in the Central Coast region. It is thereforenot surprising that a larger share of the deforested land is converted intoresidential land in this region.The structure of land use in the Central Highlands region is quite different

due to the high propensity of cash crop agriculture, primarily of coffee but alsoof rubber, tea, cocoa, and other types of cash crops. A large share of land (around50 per cent in 2014) is devoted to perennial crops while only about 30 per centis used for rice and other annual crops. There is also aminor trend of conversionof forested areas into agricultural and residential land in this region.Consistent with its nickname as ‘the rice bowl of Viet Nam’, the majority of

land in the MRD is used for rice production. In 2014, more than 60 per centof land was used for this purpose. There is no clear trend in land use sharesover time in this region. This, combined with the possibility of measurementerrors, means that the year-to-year differences in this region will not beexplored further.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

6 8 10 12 14 6 8 10 12 14 6 8 10 12 14 6 8 10 12 14 6 8 10 12 14Red River Delta North Central Coast Central Highlands Mekong River Delta

Year and Region

Rice Nonrice annuals Perennials Residential Forest

Figure 3.4 Average land use shares by year and regionSource: Author’s calculations based on VARHS database. The shares are calculated as simple averagesof commune shares. The five categories always sum to 100%. Other types of land such as watersurfaces, mountains, etc. are not included in the calculations.


56

In conclusion, both the structure of land use and the evolution of this overtime has varied between regions. Clear correlations between income levels andthe structure of land uses stand out. Most land in the two relatively prosperousdelta regions are used for rice cultivation. In the two poorest regions, theNorth and the Central Coast, a trend towards deforestation is clearly observed.This is a sign of both population increases and expansion of the economicactivities undertaken in these regions. In general, the share of land used forresidential purposes has increased, which reflects both rising incomes acrossthe country and rising population densities.

3.3 Provision of Public Goods and Infrastructure

This section investigates to what extent the high level of public investmentsand the expected expansion of infrastructure and basic services can beobserved in the communes of the VARHS. The section also investigates theheterogeneity of the expansion across regions.

The set of services offered at the commune level establish the frameworkconditions under which households work, earn income, and make decisions.A priori, we expect substantial improvements over this period: public invest-ment in infrastructure was above 10 per cent of GDP per year between 1997and 2009 (Thanh and Dapice 2009). The length of paved roads in the countryalmost quadrupled, while the number of households with access to pipedwater rose from 12 per cent in 2002 to 76 per cent in 2009 (VietnamDevelopment Report 2012). These figures document a high level of growthbut are also indicative of low initial levels of infrastructural services.

That importance of infrastructure and public goods in a transitioning econ-omy such as Viet Nam is well documented in the academic literature. Thisincludes an estimated return to education of 13 per cent per year of primaryeducation in Viet Nam (Moock, Patrinos, and Venkataraman 2003). It alsoincludes evidence of positive impacts of rural road improvements in Viet Namon primary school completion rates and on the presence and frequency ofmarkets (Mu and van de Walle 2011). Mu and van de Walle (2011) also foundthat rural road improvements led households to switch from agriculture tonon-agricultural activities, thus supporting the structural transformation pro-cess. Health is an important part of human capital, and it is therefore notsurprising that health shocks have been found to have negative effects onincome and welfare (Wagstaff 2007; see also Chapter 10 of this volume). Thisprovides a rationale for an extended network of clinics and hospitals. How-ever, it should be noted that availability of health facilities is only one part ofthe equation; the cost to the households of obtaining health care is another.Before the DoiMoi reforms, health care in Viet Namwas largely covered by the


57

state, although patients covered the costs of drugs. The Doi Moi reforms cutstate support for the health sector; government hospitals and clinics wereallowed to charge patients additional fees; and private health facilities werelegalized. This led to increased out-of-pocket expenses for households. Inrecent years, Viet Nam has moved towards a single-payer system, althoughout-of-pocket expenditures are still substantial, standing at 57 per cent in 2010(Somanathan et al. 2014).Even though the level of the commune is a natural level of analysis for

provision of infrastructure and other types of public goods, since it encom-passes the daily surroundings of the households, the decisions to providethe types of infrastructure analysed in this section are not made solely bythe commune administrators. Some facilities, such as public primary andsecondary schools and health-care centres, are managed at the district level(the second-lowest administrative level) and funded at the provincial level.Commune authorities can request provision of these facilities but they do notmake the decision. For other types of facilities, such as roads and streetlights,the funding and decision process differs depending on the type supplied.A third type of facilities, such as extension shops and centres, can also befunded at the provincial level—but private non-profit and for-profit agentsalso operate in this market. Similarly, some communes have private primaryand secondary schools. The results of the following should therefore be inter-preted as an analysis of the conditions of households living in these com-munes and not as a view into the decisions of commune authorities as towhich facilities and services to provide to the community.Figure 3.5 shows the presence in the communes of six types of facilities from

2006 to 2014. There is evidence of some improvement. For example, the shareof communes that had markets and secondary schools is significantly higherin 2014 than it was in 2006. Likewise, the share of communes that had health-care centres and a clinic or hospital was significantly higher in 2014 than in2008, the first year in which these variables were recorded. There is no signifi-cant change in the share of communes that had primary schools, but this isbecause almost all communes already had one in 2008. Likewise, there hasbeen no improvement overall in the share of communes with a post office. Infact, this share declined slightly in the sub-period 2006–12. In general,improved provision of these types of public facilities did not take place evenlythroughout the period. The presence of post offices, clinics or hospitals, andhealth-care centres all declined at some point from 2006 to 2014. This showsthat transformation, even at the somewhat aggregated commune level, is acomplex process where setbacks can occur in some aspects in some years whileothers improved. It also highlights the importance of not over-interpretingshort-term changes. A longer time horizon, such as the one of VARHS, whichcovers eight years, is needed in order to conduct meaningful inference.


58

Figure 3.6 investigates how the distances to the nearest bus stop, the nearestmain road, the district centre, and the distances to the nearest extensioncentre and extension shop have evolved. For all distances, except the distanceto a bus stop, there are indications of improvement over the period. The shareof communes that have less than 2.5 km to go to reach to these locations hasincreased. However, there is less improvement at the other end of the distri-bution: the share of communes that do not have facilities within 20 km, forexample, is largely unchanged over the period for all indicators. For a largegroup of households, distances have been reduced over the period, but themost remote households are not becomingmore integrated. This is potentiallyworrisome as it indicates that some communes are not reaping the benefits ofthe improvements that have taken place at the aggregate level over thisperiod. However, it is also possible that this is simply a snapshot of a trans-formation process that is in progress but not yet completed, and that, at thisspecific point in time, the progress of transformation is not equally far pro-gressed in all communes. If this is the case, it is possible that the communesthat are currently lagging behind will start to catch up in the near futurewithout further intervention or additional policies. Nonetheless, the hetero-geneity of the transformation process is a topic that warrants careful attentionin the coming years, and if the poorest communes do not show signs ofcatching up, additional measures should be taken to help facilitate a decreaseof the gap.

0

10

20

30

40

50

60

70

80

90

100

2006 2008 2010 2012 2014

Perc

enta

ge

of

com

mun

es

Year

Market Post office Primary school Secondary school

Health care center Clinic or hosptial

Figure 3.5 Presence of six commune facilities over time, % of communesNote: Information on primary schools, health-care centres, and clinics or hospitals is only availablefrom 2008. Information on post offices is not available for 2008 and 2010.Source: Author’s calculations based on VARHS database.


59

Figure 3.7 shows how distances varied in 2014 by region. The close proxim-ity to the urban centres of Hanoi and Ho Chi Minh City are apparent in thedistances for the Red RiverDelta and theMRD regions.Here, infrastructure tendsto be more developed and distances shorter. The North region, with its moun-tainous terrain and low population density, is the clear loser on all distancemeasures except distance to amain road. In the North,more than 20 per cent ofcommunes have more than 20 km to the nearest bus stop. This is the case forless than 5 per cent of communes in the Red River Delta. Likewise, more than30 per cent of communes in the North are more than 20 km from the districtcentre, whereas this is the case for only a few per cent of communes in the RedRiver Delta. On most indicators, the Central Highlands communes have thesecond-longest distances after the North region.Figure 3.8 shows the prevalence of streetlights and drinking water distribu-

tion networks in the communes by region and over time. The figure shows thepresence of these two types of networks but not the coverage. Coverage is less pre-cisely measured in the data, but the available information indicates that, in mostcommunes that have streetlights and water distribution, less than 50 per cent

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2008

2010

2012

2014

2008

2010

2012

2014

2008

2010

2012

2014

2008

2010

2012

2014

2008

2010

2012

2014

Bus stop District center Extension centre Extension shop Main road

<2.5 km 2.5–5 km 5–7.5 km 7.5–10 km 10–20 km >20 km

Figure 3.6 Distances to transportation and other facilities by year, % of communesNote: Distance to bus stop is measured from the People’s Committee office. All other distances aremeasured from the centre of the commune. Often, the People’s Committee office is located at thecommune centre.Source: Author’s calculations based on VARHS database.


60

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Red

Rive

r D

elta

Cen

tral

Coa

st

Cen

tral

Hig

hlan

ds

Mek

ong

Rive

r D

elta

Nor

th

Nor

th

Cen

tral

Coa

st

Cen

tral

Hig

hlan

ds

Mek

ong

Rive

r D

elta

Red

Rive

r D

elta

Cen

tral

Hig

hlan

ds

Cen

tral

Coa

st

Red

Rive

r D

elta

Mek

ong

Rive

r D

elta

Nor

th

Red

Rive

r D

elta

Cen

tral

Coa

st

Mek

ong

Rive

r D

elta

Cen

tral

Hig

hlan

ds

Nor

th

Red

Rive

r D

elta

Nor

th

Cen

tral

Coa

st

Mek

ong

Rive

r D

elta

Cen

tral

Hig

hlan

ds

Bus stop District center Extension centre Extension shop Main road

<2.5 km 2.5–5 km 5–7.5 km 7.5–10 km 10–20 km >20 km

Figure 3.7 Distances to transportation and other facilities in 2014 by region, % ofcommunesNote: See Figure 3.6, Note.Source: Author’s calculations based on VARHS database.

0

10

20

30

40

50

60

70

80

90

100

2008 2010 2012 2014 2008 2010 2012 2014Streetlights Water distribution network

Perc

enta

ge

of

com

mun

es

Year


Figure 3.8 Percentage of communes with streetlights and drinking water distributionnetwork by region and over timeSource: Author’s calculations based on VARHS database.


61

of households are placed directly on the streetlight and water distributiongrids. Both indicators show steady progression over time in most regions.The Red River Delta communes are the clear leaders in terms of streetlights.In 2014, more than 90 per cent of communes had at least some streetlights.However, the other regions, with the possible exception of the North region,have been catching up over the period. A similar picture emerges from thewater distribution network information. Here, the MRD communes are thebest off. In 2014, more than 90 per cent of communes had at least a limitedwater distribution network. Some catch-up has happened over time, exceptfor in the North region where rates appear to have fallen. Communes in theRed River Delta region have the lowest prevalence of water distribution net-works among the five regions, which may be explained by the close proximityto surface water from the delta—even though this is likely to be polluted.A third type of infrastructural network is internet access. Even though

mobile phone technology and wireless data speeds have improved at a rapidpace over the period, wired internet access is still of importance since wirelesscoverage can be patchy or non-existent in some rural areas. Many aspects ofinternet usage are also easier using an internet-connected computer ratherthan a mobile device. Finally, high-speed wireless access requires a nearbyantenna connected to an internet cable. In this sense, commune internetaccess points can be considered a proxy measure for high-speed wirelessinternet access. As Figure 3.9 shows, there has been substantial progress ininternet access over the period in all regions. In 2006, 33 per cent of com-munes had at least one internet access point. This had increased to 87 per cent

0

10

20

30

40

50

60

70

80

90

100

2006 2008 2010 2012 2014

Perc

enta

ge

of

com

mun

es

Year


Figure 3.9 Percentage of communes with at least one internet access point by regionand over timeSource: Author’s calculations based on VARHS database.


62

in 2014. The communes in the Red River Delta region were early adopters.Already in 2006, 69 per cent of communes had a connection to the Internet.Again, the North region lags behind. In 2014, only around 75 per cent ofcommunes in the North had access. Chapter 8 explores internet availability inmore detail using information at the household level.

In conclusion, it is possible to observe real improvements across a multitudeof indicators related to commune facilities and infrastructure in the period2006–14. The prevalence of commune facilities such as markets, secondaryschools, health-care centres, and clinics has increased. The high level of infra-structural investments seems to have had an effect on the ground: distances toroads and extension shops have been reduced and the existence of waterdistribution networks and internet access points has become more widespread.While some progress can be observed in all the regions of the country thatVARHS covers, stark interregional differences are present. On a few indicatorssuch as internet access, the poorest regions of the North and the Central Coastare doing aswell in 2014 as the richest provinces of the Red River andMRDwerein 2006. On many others, such as distance to crucial pieces of infrastructure,such as main roads, extension shops, and bus stops, as well as infrastructuralindicators such as streetlights and water distribution, the poorest regions were,in 2014, still very far from 2006 levels of best-off regions.

3.4 Past, Present, and Future Commune Problems

The previous sections have considered a series of objective indicators of thetransformation process observed at the local level. This penultimate sectioninstead considers subjective answers of commune administrators as to whichproblems affect their commune the most.

Figure 3.10 shows which problems commune administrators listed as hav-ing affected the commune the previous year, starting in 2010 when this set ofquestions was first asked. The questionnaire also includes questions on whichproblems commune administrators at the time thought would affect thecommune in the coming two years. The answers in 2014 to this question arealso included in the figure to give an insight into commune administrators’thoughts on which problems they thought that they would face in the nearfuture. This section of the questionnaire was expanded in 2012 to include aquestion on climate change.

The overall message is quite positive. All the problems listed in the figure,except climate change, affect a smaller share of communes inmore recent years,compared to 2010. The problem affecting the highest share of communes since2010 was natural disasters and diseases. In 2010, more than 60 per cent ofcommunes were affected by this. The importance of natural disasters anddiseases has declined slightly since then. This can be either due to changes in


63

the resilience of communes or simply due to fewer or less severe incidents in2012 and 2014 than in 2010. The secondmost widespread problem in 2010waspower, roads, and water. The share of communes whowere affected by this alsofell in both 2012 and 2014. Even fewer administrators expected this to be animportant problem in the coming two years. This lines up with earlier results,which found improvements in distances to roads as well as improvements inpower andwater distribution networks. Asmentioned earlier in this section, theonly problem that appears to be affecting more and more communes is climatechange. In 2012, 24 per cent reported that their commune was affected byclimate change. In 2014, this share increased to 27 per cent; 36 per cent expectthis to be among the most important in the coming two years. Climate changecan result in changes of the mean temperatures, mean precipitation, and so on,as well a higher frequency of extreme weather events. Using only observationsfrom one’s own commune, climate change can be hard to observe since adverseweather events such as floods, storms, and year-on-year temperature changeshappen even in the absence of climate change. It is possible that natural year-on-year variation is driving the increase in commune administrators whoworryabout climate change. Another possibility is that the results simply reflectincreased awareness of the issue and no real change in conditions on theground. Climate change is an issue that has received increased attention from

0

10

20

30

40

50

60

70

2010 2012 2014 2016

Perc

enta

ge

of

com

mun

es

Year

Natural disasters & diseases Power, roads and water Irrigation systems

Environmental problems Social evils Climate change

Figure 3.10 Percentage of communes affected by different commune problems in thepast, present, and futureNote: Respondents were asked to list all problems affecting the commune. The list of problems alsoincluded health and education, access to health and education, quality of health and education,gender discrimination and family, and ethnic discrimination. These were left out of the figure sincevery few communes chose these options. Climate change was not included as an option in 2010. In2014, respondents were asked which problems they expected to be important in the coming twoyears. These answers are included in the figure as 2016 responses.Source: Author’s calculations based on VARHS database.


64

media outlets as well as policy makers in recent years. However, one should notdismiss out of hand that, in more than a quarter of communes in the VARHSdatabase, administrators feel they are already experiencing the negative impactsof climate change, and an additional 10 per cent expect to experience sucheffects in the coming two years.

Figure 3.11 breaks the 2014 answers down by regions. There are differencesacross regions both in the share of communes that experience problems aswell as which problems are most prevalent. Less than 40 per cent of com-munes in the Red River Delta experienced natural disasters and diseases in2014. This is lower than in all other regions. The Central Coast region washit the hardest: almost 80 per cent of communes were affected by such ashock in 2014. Power, roads, and water are less important in the MRD, wherearound 30 per cent of communes experienced these issues as a problem. It is,however, quite important in the more mountainous regions of the North andthe Central Highlands where population densities are lower and there aremore complicated topographies. The Central Highlands is also the regionwhere the highest share, more than 35 per cent, of communes report thatirrigation systems are problematic. This is most likely due to the combinationof the need for proper irrigation for many of the cash crops grown in theseareas and more complicated access to water compared to the lowland anddelta regions. The Central Highlands is also the region where problems ofsocial evils are most prevalent. Social evils include, but are not necessarily

0

10

20

30

40

50

60

70

80

Natural disasters& diseases

Power, roads andwater

Irrigation systems Environmentalproblems

Social evils Climate change

Perc

enta

ge

of

com

mun

es


Figure 3.11 Percentage of communes affected by different problems in 2014 by regionSource: Author’s calculations based on VARHS database. Respondents were asked to list all problemsaffecting the commune. The list of problems also included health and education, access to healthand education, quality of health and education, gender discrimination and family, and ethnicdiscrimination. These were left out of the figure since very few communes chose these options.Climate change was not included as an option in 2010.


65

limited to, drug and alcohol abuse, prostitution, and gambling. More than45 per cent of communes in the Central Highlands have problems with thesecompared to 32 per cent at the national level. The Central Coast is also theregion where most commune administrators felt adverse effects of climatechange. More than 45 per cent of communes experienced problems related toclimate change here. In the regions of the North and the MRD, this share wasless than 20 per cent.

3.5 Conclusion

This chapter has documented the structural transformation process as it hastaken place at the commune level in rural Viet Nam over the period 2006–14.Significant change and improvements were found inmany types of indicators.However, the pace of transformation varies greatly between different regions.This is partly due to varying initial conditions in 2006 and partly due tosubstantial differences in occupational and agricultural structures, which areat least partly determined by geographical conditions.While the changes and improvements in living conditions that have taken

place over the period are substantial, the observed changes should not beoverstated. On the ground, things were the same in 2014 as they were in2006. Agriculture is still by far the most important occupation, and rice is stillthemost important agricultural crop. Instead, the picture that emerges is one ofsteady and gradual progress in many different dimensions. The occupationalstructure was more diversified in 2014 than in 2006, with more communesreporting occupations such as construction, other services, and aquaculture tobe of importance. Likewise, land use diversity has increased, with more landbeing used for residential purposes towards the end of the period.The study also documents steady improvements in the provision of

public goods and access to basic infrastructure in communes. Here, however,the regional differences are stark: the poorer and less populated regions ofthe North and Central Coast, and to some degree the Central Highlands, areworse off on a wide range of distance indicators as well as on connection to theInternet and to a water distribution network. However, on some of the indi-cators of commune facilities, the poorer regions are doing relatively betterthan the richer regions located in the deltas near the population centres ofHanoi and Ho Chi Minh City.In the provision of some infrastructural services, it appears that while many

communes experience progress on these fronts, the worst-off communes havenot been as fortunate. This means that while the economic growth of VietNam has been quite broad-based, there may be a need to do even more in thefuture in for these communes to catch up to the rest of the country.


66

Section 3.4 showed how commune problems, as experienced by communeofficials, have changed over time. This piece of evidence is quite positive. Mostproblems affected fewer communes in 2014 than in 2006. However, there hasbeen an increase in the number of communes that see climate change as aproblem, and the number of communes that expect this to be a problem in2016 is even higher. Climate change is a problem that is unsolvable at thecommune level and even at the national level. What can be done at bothlevels is to help farmers and other households to adapt to climate changes.These matters should receive increased attention in the coming years anddecades.

References

Moock, P. R., H. A. Patrinos, andM. Venkataraman (2003). ‘Education and Earnings in aTransition Economy: The Case of Vietnam’. Economics of Education Review, 22(5):503–10.

Mu, R. and D. van de Walle (2011). ‘Rural Roads and Local Market Development inVietnam’. Journal of Development Studies, 47(5): 709–34.

Somanathan, A., A. Tandon, H. L. Dao, K. L. Hurt, and H. L. Fuenzalida-Puelma (2014).Moving Toward Universal Coverage of Social Health Insurance in Vietnam: Assessment andOptions. Directions in Development: Human Development. Washington DC: World BankGroup.

Thanh, X. N. and D. Dapice (2009). Vietnam’s Infrastructure Constraints. Policy DialoguePaper 3. Cambridge MA: ASH Institute for Democratic Governance and Innovation,John F. Kennedy School of Government.

Vietnam Development Report (2012). Market Economy for a Middle-income Vietnam.Washington DC: World Bank.

Wagstaff, Adam (2007). Health Insurance for the Poor: Initial Impacts of Vietnam’s HealthCare Fund for the Poor. Policy Research Working Paper Series 4134, Washington DC:World Bank.

World Bank (2012). Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on PovertyReduction and the Emerging Challenges. Washington DC: World Bank.


67

4

Commercialization in Agriculture, 2006–14

Chiara Cazzuffi, Andy McKay, and Emilie Perge

4.1 Introduction

Agriculture has always been the predominant production activity in ruralViet Nam. Today it remains a major activity in almost all rural areas, eventhough there has been a substantial development of non-farm activities inmany such areas more recently. Without doubt, a key part of Viet Nam’seconomic transformation over the past thirty years has been the substantialprogress made in agriculture. Data from the Food and Agricultural Organ-ization (FAO) show that production of the country’s dominant crop, rice,increased almost every year between 1975 and 2013, and increased by afactor of over four during this period (Figure 4.1). Starting off as a netimporter of rice, the country switched, in the 1980s, to being a substantialexporter of rice, and is now the second biggest rice exporter in the worldaccording to the International Rice Research Institute (IRRI). Over this timeperiod, Viet Nam has also become much more involved in the cultivation ofcash crops, notably coffee, and fishery products are also an area of substan-tial export growth.Agriculture in Viet Nam has changed substantially since the reunification of

the country in 1975, particularly following the launch of the Doi Moi reformprogramme in 1986. Following reunification, agricultural production con-tinued to be organized on a collective basis. Viet Nam was a very closed andhighly controlled economy at that time, but this was also a period whenagricultural performance was poor. The country experienced substantialfood shortages, if not famine, at different periods in the 1970s and 1980s.This led the government to develop a major focus on food security, a concernwhich remains to the present day, as evident in the government’s activeresponse to the 2007/8 food crisis (McKay and Tarp 2015).

A first gentle step towards liberalization was the introduction of a productcontract system in 1981, which allowed some cooperatives to give individualhouseholds access to some plots of land, though still requiring them to selltheir produce to the cooperative (Marsh, MacAulay, and Van Hung 2006). Butmuch more substantial reforms came into place following the 1986 Doi Moireforms. At this point, the need for giving a much greater role to individualhouseholds and the private sector was recognized. Resolution No. 10, passedin 1988, gave households greater production rights, including the right tochoose their own buyers.

A series of land reforms were then introduced. In 1988 land was dividedamong households, who were given fifteen-year leases (Glewwe 2004; McCaigand Pavcnik 2013). The 1993 Land Law then gave farmers land use rights, nowwith tenure of twenty years (fifty years for perennial land). This includedrights to transfer, exchange, lease, inherit, and mortgage their land, and sogave farmers greater security. Land titling was introduced at this time andimplemented relatively quickly, such that, by 1997, half of land had beentitled (Benjamin and Brandt 2004; McCaig and Pavcnik 2013). Farmers wereallocated with land-use certificates (LUC) or ‘red books’, generally in thenames of both the household head and spouse. A series of further revisionswere made in 1998, 1999, and 2001, after which a new Land Law was intro-duced in 2003 (Marsh, MacAulay, and Van Hung 2006; Markussen, Tarp, andvan den Broeck 2011). But, at the same time, the authorities, generally atcommune level, continue to place significant restrictions on the use of land,in particular requiring the cultivation of rice on many plots. Much of this ismotivated by concern with food security, given that rice is also the staple

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

1975 1979 1983 1987 1991 1995 1999 2003 2007 2011

Tho

usan

ds

of

ton

nes

Year

Figure 4.1 Rice production in Viet Nam, 1975–2014 (tonnes)Source: Data from FAOSTAT <http://faostat3.fao.org/home/E>.


69

http://faostat3.fao.org/home/E

consumption commodity in Viet Nam. On non-restricted plots, though,households are able to decide what to cultivate.Restrictions on internal and external trade of agricultural outputs and

inputs were gradually relaxed over this period. In particular, quotas wereused to limit rice exports, motivated by food security concerns, but thesewere significantly relaxed over the 1990s. Restrictions were also placed oninternal trade at this time and on imports of fertilizers, but again, over time,these were relaxed. As a result, households became free to sell their output asthey choose to, and generally had access to production inputs. Rice product-ivity increased substantially, for instance from 3.3 tonnes per hectare in 1992to 4.9 in 2006 (Benjamin et al. 2009; McCaig and Pavcnik 2013). Increasingly,over time, households who produced sufficient quantities were generally sell-ing to private traders or enterprises, while those producing smaller quantitiesoften sold to other households when they had a surplus. The private sector iscertainly the dominant buyer of rice in rural Viet Nam.The Viet Nam Access to Resources Household Survey (VARHS) enables a

detailed analysis of the roles played by households in different agriculturalactivities, and the panel feature of the data allows the dynamics of this role tobe investigated over the period 2006 to 2014. The largemajority of householdsinterviewed in VARHS earn at least some of their income from agriculture,even if, over time, non-agricultural livelihoods are becoming increasinglymore important, as expected with economic development. Although, in sev-eral provinces, wage earnings have overtaken agriculture as themain source ofincome (see also Chapter 10), most households still have some income fromagriculture or natural-resource-based activities. In provinces in the CentralHighlands and Northern Uplands, agriculture remains the dominant activity.The VARHS collects detailed information on the agricultural activities

undertaken by households: the crops grown and sold, livestock activities,land use, including engagement in aquaculture, and use of inputs, amongother things. This data enables a substantial and detailed analysis of theseissues. This chapter is only a start at this, presenting an initial analysis, lookingat households’ engagement in three important areas of activity: rice cultiva-tion, production of cash crops, and engagement in household-level aquacul-ture activities. Since most households grow rice, the chapter focuses oncommercialization, seen through sale of rice. But it also considers coffeeproduction, a key cash crop in Viet Nam, as well as household engagementin aquacultural activities on their own land.Again, the analysis is based on the 2,162 households included in the bal-

anced 5-wave panel between 2006 and 2014, looking in particular at theextent to which households cultivating rice sell it and on what scale. Formost households engaged in agriculture, rice commercialization is a dominantincome source. For cash crops and aquaculture, the outputs are sold almost by


70

definition. The interest here is on modelling the correlates of householdsengaging in these activities. Part of the analysis here compares the five cross-sections that make up the panel dataset, but we also exploit the panel featureof the data to look at the dynamics of these activities over time.

The rest of this chapter is structured as follows. Section 4.2 briefly reviewssome relevant international literature, after which Section 4.3 provides anintroduction to the extent of household participation in these activities.Section 4.4 examines patterns of engagement in rice cultivation and sales,cash crops, and aquaculture by geographic region and income quintile, follow-ing which Section 4.5 exploits the panel to examine, among other things, theextent of consistency of these activities over time at the household level. Aneconometric analysis of correlates of engagement in these different commercialactivities is presented in Section 4.6, after which Section 4.7 concludes.

4.2 Some Relevant Literature

This chapter relates to the growing literature that examines the determinantsof small farmer participation in commercial activities in agrarian economies.Much of this literature focuses on food crops, which households often prod-uce for their own consumption but may also choose to sell. This literature hassought to understand primarily the role of transaction costs and marketfailures in smallholder decision-making. Differential asset endowments,together with differential access to public goods and services that facilitatemarket participation, are identified as key factors underlying heterogeneousmarket participation among smallholders (Key, Sadoulet, and de Janvry 2000;Barrett 2008). Differences in transaction costs across households are alsoimportant determinants of market participation: each household faces somefixed time and monetary costs in searching for available marketing options,and, if high enough, these costs, invariant of the quantity transacted, mayprevent market participation altogether. According to Goetz (1992), transac-tion costs affect market participation behaviour through the labour–leisurechoice: thin markets make it costly (i.e. time consuming) to discover tradingopportunities. Similarly, poor market access, due to lack of transport, distance,and/or barriers such as ethnicity or language, increases households’ cost ofobserving market prices in order to make transaction decisions, thus reducinghouseholds’ leisure time (Goetz 1992).

For staple food markets in particular, another important factor influencingthe participation decision is risk, and household attitudes towards it. House-holds concerned about their own food security and facing a high degree ofprice and non-price risk, especially in the presence of missing or imperfectcredit and insurance markets, may choose not to sell in the attempt to ensure


71

that their own consumption requirements can be met. On the other hand,lack of liquidity resulting from an absence of alternative income sources andcredit may sometimes also force households to sell rice to generate cash inorder to meet other non-food expenditures.The determinants of smallholder participation in agricultural markets have

been investigated empirically frequently in the context of sub-Saharan Africa.These studies identify strong positive associations between market participa-tion and: (1) household income and assets (especially land, but also livestock,labour, and equipment);1 (2) access to credit and insurance;2 (3) input use andaccess to extension services;3 and (4) low levels of transaction costs, includingtransport costs and information costs.4

The literature on aquaculture is significantly less developed than it is inrelation to the selling of food crops or the choice to engage in cash cropproduction, but similar factors are likely to be relevant here as in the case ofcash crops.With respect to Viet Nam, Rios, Shively, and Masters (2009) find that house-

holdswithhigher productivity tend toparticipate in agriculturalmarkets regard-less of market access factors (e.g. distance to roads or quality of transportnetworks). Such a finding suggests that programmes targeted at improvingpoorer households’ productive capital, and other assets, have the potential toincrease both productivity andmarket participation, while investments inmar-ket access infrastructure seem to be relatively less of a priority (Rios, Shively, andMasters 2009). It seems then, that even in the early 1990s, Viet Nam had muchbetter coverage of basic rural infrastructure in most regions compared to coun-tries with similar levels of income (Aksoy and Isik-Dikmelik 2007).The analysis in this chapter draws and builds on prior analyses by the authors

of some of these issues based on earlier waves of VARHS (McCoy, McKay, andPerge 2010; Cazzuffi et al. 2011; Cazzuffi and McKay 2012). However, thesestudies also addressedmore detailed issues not considered here for lack of space,for example the channels used to sell rice (Cazzuffi and McKay 2012) or theanalysis of open access fisheries (McCoy, McKay, and Perge 2010).

4.3 Agricultural Activities in the VARHS Panel

In the VARHS panel dataset, 100 per cent of households in 2006 reportedincome from crops, livestock, or aquaculture as one or more sources. This

1 Nyoro, Kiiru, and Jayne (1999); Cadot, Dutoit, and Olarreaga (2006), Stephens and Barrett(2006); Boughton et al. (2007); Levinsohn and McMillan (2007).

2 Cadot et al. (2006); Stephens and Barrett (2006). 3 Alene et al. (2008).4 Heltberg and Tarp (2002); Alene et al. (2008); Ouma et al. (2010).


72

proportion fell gradually over time, but in 2014, still 86.4 per cent of house-holds reported positive income from one or more of these sources. Thisreinforces the point made in Section 4.1 about the importance of agriculturalor aquaculture activities for almost all households.

Figure 4.2 reports some summary statistics relating to these three activitiesfor households included in the five-wave panel, treating the different waves asseparate cross-sections for now. A large majority of households grows rice ineach of the years. The proportion does decline gradually over time, but evenby 2014 more than 65 per cent of households grow rice in at least one of theirplots. As well as many households having been required by crop restrictions togrow rice, most locations covered by the survey are very suitable for ricecultivation.

The next set of columns in Figure 4.2 relates to the proportion of rice-growing households that sell some of their output. Starting out from nearly50 per cent of households in 2006, participation in rice sales shows a consist-ently increasing trend over time. While the number of households growingrice may have decreased between 2006 and 2014, an increasing proportion ofthese are selling. This latter effect outweighs the former, such that the absolutenumbers who sell show an increase. The survey also reports on channels ofsales, the most important channels being sales to traders or sales to otherindividuals or households. Channels vary by province and, unsurprisingly,the scale of sales reflects the channel used. The following set of columns ofFigure 4.2 report on the average proportion of the harvest sold, which again

0

10

20

30

40

50

60

70

80

90

Growing rice Selling rice Harvest sold Growing cashcrops

Doing aquaculture

Perc

enta

ge

2006 2008 2010 2012 2014

Figure 4.2 Some summary characteristics relating to commercialization for the fullsampleSource: Authors’ computations based on VARHS data for years 2006 to 2014.


73

shows an upward trend over time. The first years of the panel were a periodwhere the rice price increased significantly, but the extent of commercializationaccording to these two indicators has continued to increase since, even thoughthe rice price has subsequently fallen. This increasing commercialization takesplace alongside continued increases in rural household income over this period(see McKay and Tarp 2015).The remaining groups of columns in Figure 4.2 relate to the extent of

household engagement in cash crop production and in aquaculture activitieson their own land. A small minority of households participate in these activ-ities and, in the case of aquaculture at least, there may be a declining trend.But the choice to undertake these activities is a significant investment byhouseholds; furthermore, climatic and other conditions also need to be appro-priate. The dominant cash crop cultivated by households in the survey iscoffee, which is grown predominantly in the Central Highlands provinces.Other cash crops, grown on a smaller scale, include tea, cocoa, cashew nut,sugarcane, pepper, and rubber. Around 10 per cent of households earn someincome from aquaculture, which requires households to convert one or moreof their plots into a pond; this can also be a relatively labour-intensive activityand one with an uncertain return from one year to another.What is clear from this initial introductory analysis is the importance of

agricultural activity, especially rice, for households, and the extent of engage-ment in sales for a majority of these households. That in itself is a signal of thesuccess with which these activities have been conducted in rural Viet Nam.However, the analysis to date is only conducted at an aggregate level and doesnot exploit the panel features of the dataset; the remainder of this chapter nowanalyses these three activities separately and in more detail.

4.4 Rice Cultivation and Sales, Cash Crops, and Aquaculturein Rural Viet Nam

While the role of rice as a dominant crop in Viet Nam has been stressed inSection 4.3, Table 4.1 shows variations in its importance by province and byhousehold income quintile. The numbers of households cultivating rice arevery high in the three Northern Upland provinces (Lai Chau, Dien Bien, andLao Cai) and do not fall over time. Typically, 90 per cent or more of house-holds grow rice there. By contrast, in Dak Nong and Lam Dong in the CentralHighlands and in Khanh Hoa in the South Central Coast, relatively fewhouseholds grow rice. The remaining provinces lie between these extremes.In a number of these provinces, such as Ha Tay and Quang Nam, the propor-tion of households growing rice is falling over time. In these locations, non-agricultural activities, notably wage work, become increasingly important


74

over time (see McKay and Tarp 2015). Examining this by quintiles of percapita food consumption, it is clear that rice cultivation is higher in lowerquintiles than in higher ones, though the numbers cultivating rice remainsubstantial in the fifth quintile. In the higher income quintiles, more non-agricultural opportunities exist, reflecting both their more urbanized natureand higher levels of development. To some extent, the quintile pattern cor-relates with the geographic pattern: the Northern Upland provinces referredto here are disproportionately found in the lower income quintiles.

The VARHS data can also be used to estimate productivity in rice produc-tion, by dividing total production by the area cultivated. This gives averagefigures of 5.9 tonnes/ha in 2006, 5.4 in 2008, 6.0 in 2010, 5.2 in 2012, and 6.5in 2014. These figures show variability from one year to another, which is notsurprising. It is difficult to draw any strong conclusions about trends over theperiod on this basis, particularly as yield inevitably fluctuates from one year toanother. On a longer time horizon and with aggregate national data, Abbottand colleagues (forthcoming) identified an increase in agricultural productiv-ity between 2000 and 2011; that is not inconsistent with these yield data.But the more important point from these survey data is that these figures arevery similar to both the rice yield figure for Viet Nam reported by the IRRI in2008, which was 5.2 tonnes/ha, and to the figures quoted in Section 4.1. Theselatter figures are national while the VARHS figures relate to just these twelve

Table 4.1 Proportion of households growing rice, by year, province, and quintile

2006 2008 2010 2012 2014

ProvinceHa Tay 0.864 0.815 0.768 0.706 0.689Lao Cai 0.906 0.882 0.859 0.871 0.906Phu Tho 0.889 0.791 0.731 0.710 0.737Lai Chau 0.945 0.908 0.881 0.881 0.908Dien Bien 0.980 0.960 0.939 0.939 0.960Nghe An 0.739 0.707 0.670 0.681 0.644Quang Nam 0.824 0.820 0.784 0.734 0.694Khanh Hoa 0.417 0.236 0.389 0.361 0.361Dak Lak 0.542 0.489 0.550 0.527 0.473Dak Nong 0.380 0.283 0.250 0.293 0.293Lam Dong 0.250 0.281 0.266 0.250 0.172Long An 0.668 0.585 0.567 0.588 0.581Consumption quintile1 0.880 0.822 0.680 0.851 0.8502 0.770 0.768 0.742 0.762 0.7623 0.724 0.730 0.751 0.711 0.7294 0.598 0.583 0.718 0.651 0.6255 0.481 0.454 0.576 0.459 0.494Total 0.764 0.710 0.685 0.666 0.654

Source: Authors’ computations based on VARHS data for years 2006 to 2014.


75

provinces, but the very close nature of the estimates provides reassuranceabout the reliability of the data collected on rice cultivation in VARHS.The geographic disaggregation of the proportion of rice growers selling their

output is presented in Table 4.2, showing the very highmarket engagement inthe Long An province in particular. While between 55 and 70 per cent of ruralhouseholds in this province grow rice, almost all of them sell. The Long Anprovince is very much a commercial heartland of Viet Nam.Many householdsthere grow and sell on quite a large scale and they have themajor advantage ofbeing very close and well connected to a highly concentrated population inand around Ho Chi Minh City. When these households choose to grow ricethey almost all aim to sell, including to exporters, and as can be seen inTable 4.3, they also sell by far the highest proportion of their output.Rates of sales are much lower in other provinces, not least in the Northern

Uplands provinces, where most households grow rice. It is clear that many ofthese households are not able to produce enough to be able to sell on aconsistent basis; they also have significantly greater difficulty in getting accessto buyers. A similar point is true of Phu Tho, where again many householdsgrow rice. Although this province has much easier access to Hanoi and biggerurban centres than the Northern Uplands provinces, still relatively few ricegrowers sell. This clearly reflects the scale of production plus potentially alsothe importance of market access to participate in rice commercialization.Among the other provinces, Quang Nam, Dak Lak, and Khanh Hoa haverelatively high proportions of rice growers engaged in sales.The geographic distribution of the proportion of output sold is also shown

in Table 4.3. In almost all provinces, except Long An, households are selling aminority, and often a small minority, of their output. It is quite clear that ricecultivation and commercialization is radically different in Long An compared

Table 4.2 Proportion of rice-growing households who sell, by province and year

2006 2008 2010 2012 2014

Ha Tay 0.424 0.512 0.568 0.605 0.590Lao Cai 0.545 0.467 0.562 0.662 0.506Phu Tho 0.212 0.374 0.350 0.218 0.324Lai Chau 0.515 0.364 0.302 0.479 0.515Dien Bien 0.887 0.411 0.505 0.581 0.484Nghe An 0.518 0.459 0.341 0.484 0.645Quang Nam 0.459 0.640 0.789 0.637 0.658Khanh Hoa 0.600 0.706 0.536 0.846 0.654Dak Lak 0.465 0.625 0.472 0.609 0.597Dak Nong 0.571 0.538 0.609 0.444 0.444Lam Dong 0.250 0.611 0.529 0.813 0.455Long An 0.870 0.914 0.879 0.914 0.907



76

to the other provinces. The proportions sold are particularly low in the North-ern Uplands provinces as well as in Phu Tho, Dak Nong, and LamDong; this ispartly accounted by a relatively small scale of production. The proportion ofoutput sold generally increases with the income quintile, though again thispartly reflects the geographic distribution of the provinces, with Long Anbeing disproportionately represented in higher quintiles. Of course, bothrice cultivation and sales can fluctuate from one year to the next, an issuethat will be explored in Section 4.5, using the panel.

Tables 4.4 and 4.5 report on the percentage of households engaged in cashcrops and in aquaculture by province and quintile, and these tables againshow some quite distinct patterns. In particular, cash crops are predominantlygrown in the Central Highlands provinces of Dak Lak, Dak Nong, andLam Dong, with the dominant part of this being coffee cultivation. Cashcrop cultivation is much lower elsewhere, and almost non-existent in theprovinces of Ha Tay, Dien Bien, Quang Nam, and Long An. In general,households in higher quintiles are more likely to be engaged in cash cropcultivation even if the relationship is less strong in 2014.

The highest incidence of aquaculture is observed in the Dien Bien province;depending on the year, between one-third and one-half of households reportincome from this activity. Reasonable numbers of households in Lao Cai, PhuTho, and Long An also report earnings from aquaculture. Elsewhere the pro-portions are lower.

Table 4.3 Average proportion of rice output sold, by location, quintile, and year

2006 2008 2010 2012 2014




77

Table 4.4 Proportion of households growing one or more cash crop, by province, quintile,and year

2006 2008 2010 2012 2014



Table 4.5 Proportion of households engaged in aquaculture activity on their own land, byprovince, quintile, and year

2006 2008 2010 2012 2014

ProvinceHa Tay 0.055 0.062 0.038 0.040 0.055Lao Cai 0.294 0.176 0.259 0.212 0.176Phu Tho 0.273 0.175 0.152 0.121 0.128Lai Chau 0.156 0.101 0.037 0.037 0.028Dien Bien 0.333 0.475 0.515 0.475 0.485Nghe An 0.133 0.080 0.080 0.074 0.053Quang Nam 0.036 0.022 0.018 0.025 0.007Khanh Hoa 0.014 0.042 0.056 0.042 0.014Dak Lak 0.084 0.115 0.115 0.076 0.023Dak Nong 0.196 0.098 0.109 0.087 0.076Lam Dong 0.078 0.047 0.016 0.047 0.031Long An 0.271 0.217 0.343 0.090 0.134Income quintile1 0.141 0.123 0.118 0.095 0.0862 0.116 0.123 0.12 0.079 0.0813 0.155 0.132 0.127 0.09 0.0934 0.179 0.118 0.15 0.102 0.095 0.167 0.118 0.144 0.083 0.095Total 0.151 0.123 0.132 0.090 0.089



78

4.5 Using the Panel to Look at Production and Sales Dynamics

To date, the analysis has been entirely based on comparisons between therepeated cross-sections in the panel dataset, but looking at dynamics helpsidentify the extent to which behaviour varies over time or is consistent fromone period to another. The panel data are exploited here by looking at theextent to which households engage in these activities, growing rice, sellingrice, growing cash crops, and earning from aquaculture, in all years (Table 4.6).Those not engaged in these activities in any of the five years are also includedin these data. The patterns of those growing rice in the panel vary by provinceand quintile in much the same way as in the cross sections; and in mostlocations those that grow rice do so consistently year on year.

In relation to rice sales, Long An has by far the highest number of house-holds who sell each year in the panel. Not only domany households there sell,and sell a high proportion of their output, they also tend to do so every year.The number of consistent sellers is much smaller elsewhere, although in partthis also reflects the lower numbers of people selling in any of the cross-sections.

The numbers that consistently grow cash crops are not much lower than thenumbers reported in the cross-section. This reflects the fact that many of these

Table 4.6 Proportion of households engaged in some commercial agricultural activities inall years of the panel

Grow rice inall years

Selling in allyears*

Cash crops inall years

Aquaculture inall years

Ha Tay 0.619 0.189 0.000 0.011Lao Cai 0.776 0.121 0.071 0.047Phu Tho 0.636 0.021 0.027 0.024Lai Chau 0.853 0.097 0.000 0.009Dien Bien 0.929 0.163 0.000 0.192Nghe An 0.606 0.175 0.059 0.011Quang Nam 0.640 0.292 0.000 0.004Khanh Hoa 0.167 0.250 0.042 0.000Dak Lak 0.336 0.364 0.519 0.000Dak Nong 0.163 0.267 0.500 0.011Lam Dong 0.141 0.222 0.641 0.000Long An 0.455 0.762 0.000 0.025Quintile1 0.717 0.182 0.050 0.0302 0.559 0.289 0.091 0.0213 0.533 0.283 0.100 0.0194 0.337 0.273 0.096 0.0165 0.253 0.225 0.196 0.023Total 0.568 0.231 0.085 0.022

Note: * from among those growing each year.



79

cash crops are tree crops and therefore a long-term commitment. As in thecross-section, the numbers are highest by far in the Central Highlands prov-inces. On the other hand, the same is not true for aquaculture; the numberswith consistent earnings from this activity are consistently lower than thenumbers in the cross-section, suggesting that there is quite a lot of variabilityfrom one wave to the next. This may reflect households starting and stoppingthe activity, but it may also reflect major shocks in particular years, leading toa loss of earnings from this source.

4.6 In-Depth Analysis of Determinants of Commercialization

Three different forms of commercialization have been considered in thischapter: the choice by a household to sell some of the rice it produces, thechoice to grow cash crops, and the choice to engage in aquaculture. Someinitial descriptive analysis of the types of patterns of commercialization bylocation and income quintile have been presented in Section 4.5, but here weturn to a more detailed analysis of the characteristics of households choosingto participate in these forms of commercialization. This starts with furtherdescriptive analysis but then progresses tomultivariate analysis of the decisionby rice-growing households to sell some of their output. Following this, wepresent a briefer but similar analysis of the factors associated with householdsgrowing cash crops or engaging in aquaculture activities.Comparing rice growers who sell and those who do not (Table 4.7), the

striking difference between them is that the former cultivate larger areas ofland, spend much more on inputs, and are less likely to be poor, according tothe Ministry of Labour, Invalids and Social Affairs (MOLISA) classification.These differences are true in every year. Unsurprisingly, those householdsselling rice report much more agricultural income but not necessarily muchhigher income overall. Interestingly, those selling rice are further away fromroads on average, though this does not stop them selling; many householdssell to traders. Other differences such as household characteristics, groupmembership, and use of other inputs are much less striking or are less consist-ent across the different waves.A similar comparison of those cultivating to cash crops with those that do

not (McKay, Cazzuffi and Perge 2015: table 8) shows that households culti-vating these crops have substantially higher incomes (and agriculturalincomes) on average than those who do not, although, interestingly, theyare not systematically any less likely to be poor. It is clear that some house-holds benefit substantially from growing cash crops, but many others do not.Those growing cash crops cultivate much bigger areas on average, spendmuchmore on inputs overall (although less on inputs specifically for rice), and are


80

Table 4.7 Characteristics of households engaged in selling rice, compared to non-sellers

2006 2008 2010 2012 2014

No Yes No Yes No Yes No Yes No Yes

Total income 22,600.1 22,942.3 38,701.2 40,520.5 69,721.5 74,741.5 72,366.1 71,252.6 85,629.0 92,732.4Agriculturalincome

6,296.5 9,254.9 11,478.6 17,006.0 17,589.5 23,025.6 18,972.6 24,057.0 20,483.8 30,691.4

If poor (MOLISA) 0.259 0.222 0.242 0.183 0.170 0.121 0.230 0.154 0.160 0.104Cultivated area 7,663.4 13,175.3 6,704.2 10,451.2 6,930.5 10,173.8 6,887.9 9,909.8 6,127.3 9,997.7Cropland area 4,724.7 9,983.4 4,481.1 8,373.9 5,032.7 7,667.2 4,988.0 8,256.6 4,700.1 8,189.5Crop inputexpenses

2,529.8 6,966.8 8,043.5 25,509.3 10,517.6 28,529.5 15,741.0 38,496.6 16,310.5 41,525.4

Rice input expenses 1,284.6 5,986.4 1,972.7 10,135.2 2,459.8 10,858.7 3,804.6 14,110.0 3,638.1 15,218.0Per cent irrigated 0.705 0.766 0.711 0.841 0.745 0.856 0.801 0.879 0.193 0.182Per cent withrestrictions

0.583 0.574 0.534 0.569 0.378 0.399 0.627 0.615 0.395 0.334

If received credit 0.642 0.713 0.457 0.473 0.464 0.549 0.403 0.425 0.358 0.373If has red book 0.913 0.921 0.874 0.874 0.785 0.833 0.879 0.920 0.887 0.934Household size 4.7 4.8 4.8 4.7 4.6 4.5 4.5 4.4 4.4 4.3If Kinh 0.806 0.735 0.718 0.787 0.695 0.795 0.723 0.772 0.681 0.784If speak Vietnamese 0.977 0.961 0.969 0.965 0.979 0.994 0.987 0.988 0.990 0.995If head male 0.823 0.839 0.824 0.817 0.831 0.819 0.808 0.811 0.797 0.806Age 50.2 49.9 50.5 51.6 51.7 52.5 53.4 53.5 54.6 54.7Literacy 0.903 0.886 0.894 0.910 0.893 0.920 0.896 0.916 0.879 0.907Distance to road 0.948 1.795 3.262 12.854 2.722 2.969 2.553 3.248 1.586 2.397If has owntransport

0.883 0.875 0.913 0.940 0.901 0.947 0.912 0.947 0.578 0.632

If used extension 0.367 0.415 0.042 0.035 0.522 0.547 0.533 0.646 0.555 0.655If in farmer group 0.549 0.523 0.385 0.426 0.517 0.447 0.524 0.523 0.506 0.521If in women’sgroup

0.719 0.653 0.587 0.598 0.641 0.634 0.687 0.631 0.661 0.605

Source: Authors’ computations based on VARHS survey data for years 2006 to 2014.

more likely to have accessed credit. But in other respects there are not manyother systematic differences between cash-crop-growing farmers and thosenot cultivating cash crops. For aquaculture (McKay, Cazzuffi, and Perge 2015:table 9), those engaged in this activity earn more from agriculture (whichincludes aquaculture) and more income overall, they also cultivate largerareas. They also spend more on rice inputs, showing that many householdscombine aquaculture with rice cultivation. In addition, those engaged inaquaculture are more likely to have borrowed. In other respects, the differencesamong households engaged in aquaculture activities are less apparent.We turn now to modelling the determinants of these different activities—

selling rice (for those producing), cultivating cash crops, and engagement inaquaculture—exploiting the balanced panel dataset. In each case the out-comes are zero-one variables; these outcomes for the panel households aremodelled as a function of household characteristics in the same time period,using either a linear probability model or a probit model. Table 4.8 shows theoutcomes for the agricultural variables. The first model is estimated based onthe pooled dataset and including district-level fixed effects. The second andthird model exploit explicitly the panel features, the former using a linearprobability model and household-level fixed effects, and the latter a probitmodel with household-level random effects as well as province-level fixedeffects. As some of the explanatory variables here are liable to be endogenous,these models should be interpreted in terms of showing association ratherthan causality.In each of the three models, households which sell rice are likely to cultivate

larger areas, are likely to havemore of their area irrigated, are likely to use hybridseed and to hire more labour, are more likely to have received extensionsupport, and are more likely to have a market in their commune. Unsurpris-ingly, larger and poor households are less likely to sell, but more surprisingly,Kinh households and households who speak Vietnamese are also less likely tosell in thesemodels. Apart from the last findings, these results are quite intuitivein terms of explaining who is likely to sell rice in Viet Nam. These results arerelatively consistent across the three modelling approaches.Fewer strong correlates are identified for the likelihood of growing cash crops

(last three columns of Table 4.8). There is a strong geographic effect here, withcash crops being cultivated much more in the Central Highlands provinces(coffee especially) compared to the others included in the VARHS sample.Larger land area is associated with a higher likelihood to grow cash crops, andthose growing cash crops are much more likely to have received credit (thoughthis may be precisely a consequence of them choosing to grow cash crops). Thethird model shows a positive association between being of Kinh ethnicity andgrowing cash crops, though this is not evident in the othermodels. And there isweak evidence that those with their own means of transport are more likely to


82

Table 4.8 Regression results for correlates of selling rice and growing cash crops

Sales of rice Cultivation of cash crops

Pooled OLSwith districtFE

HouseholdFE panelmodel

RE probitmodel



RE probitmodel

coef/t coef/t coef/t coef/t coef/t coef/t

Percentage of land areawith restrictions

0.032** 0.004 0.045 0.022*** �0.001 0.083

(2.305) (0.308) (0.989) (3.570) (�0.242) (0.741)Total land area 0.003 0.004 0.006 0.003 0.003* 0.098***

(0.790) (0.973) (0.427) (1.438) (1.736) (2.875)Percentage of land areairrigated

0.054*** 0.030* 0.139*** 0.023*** 0.001 0.085

(2.817) (1.899) (2.698) (2.658) (0.241) (0.691)Area used for crop cultivation 0.022*** 0.009 0.079*** 0.009*** 0.000 0.014

(3.107) (1.081) (2.674) (2.791) (0.056) (0.259)If household received credit 0.008 0.012 0.042 0.022*** 0.014*** 0.394***

(0.676) (0.967) (1.064) (4.204) (2.932) (3.929)If household has red bookfor its land

0.029 �0.006 0.062 0.009 �0.024*** �0.365**

(1.492) (�0.255) (0.958) (0.997) (�2.742) (�2.029)Distance to nearest road 0.000 0.000 0.004 0.000 �0.000 0.005

(1.121) (0.375) (1.360) (0.593) (�0.053) (0.639)If market in commune 0.046*** 0.079*** 0.160*** �0.005 �0.006 �0.124

(3.472) (5.277) (3.649) (�0.808) (�1.084) (�1.038)If uses hybrid seed 0.023*** 0.012** 0.073*** �0.005* 0.002 0.020

(4.433) (2.524) (4.167) (�1.934) (1.012) (0.470)Expenses on crop inputs �0.002 0.002 0.046*** 0.001 0.000 0.026

(�1.181) (0.987) (3.949) (1.153) (0.238) (1.470)Amount of hired labour 0.095*** 0.070** 0.424*** 0.005 0.008 0.258

(3.710) (2.537) (3.158) (0.671) (1.180) (1.601)If household does wage work 0.018 0.004 0.049 �0.014*** �0.000 �0.077

(1.528) (0.284) (1.153) (�2.636) (�0.036) (�0.708)Household size �0.010*** �0.017*** �0.053*** 0.000 0.001 0.019

(�2.803) (�3.054) (�3.785) (0.199) (0.362) (0.479)If member of farmers’ union 0.014 �0.003 0.041 0.006 �0.004 �0.075

(1.187) (�0.254) (0.954) (1.088) (�0.725) (�0.683)If member of women’s union �0.004 �0.007 �0.013 �0.011** �0.006 �0.188

(�0.299) (�0.522) (�0.285) (�1.963) (�1.163) (�1.630)If received extension 0.035*** 0.012 0.087** 0.006 �0.002 �0.068

(2.821) (1.025) (2.259) (1.131) (�0.391) (�0.730)If owns a radio �0.003 �0.015 �0.029 0.011 0.006 0.125

(�0.172) (�0.913) (�0.553) (1.633) (1.019) (0.993)If has own transport 0.035* 0.021 0.111* 0.015* 0.012* 0.259*

(1.932) (1.143) (1.913) (1.835) (1.828) (1.769)If of Kinh ethnicity �0.082*** �0.123* �0.318*** 0.053*** 0.008 0.316

(�2.732) (�1.762) (�3.532) (4.225) (0.291) (1.401)If female headed 0.028* 0.008 0.043 �0.005 �0.017 0.084

(1.818) (0.209) (0.652) (�0.632) (�1.207) (0.395)If speaks Vietnamese �0.121*** �0.106** �0.346** 0.030 0.025 0.998*

(�2.779) (�2.059) (�2.272) (1.500) (1.260) (1.709)If poor (MOLISA) �0.042*** �0.058*** �0.156*** 0.001 0.009 0.110

(�2.762) (�3.155) (�2.915) (0.076) (1.353) (0.794)Age 0.000 0.002 0.001 �0.000 �0.001** �0.011*

(continued )


83

grow cash crops, a factor which also had a similarly weakly significant influencein the previous models on the likelihood of selling rice.Equivalent models for the likelihood of engaging in aquaculture are pre-

sented in Table 4.9. Here land area, receipt of credit, being literate, and beinga male-headed household are all positively associated with engagement inaquaculture. Larger households are also significantly more likely to undertakeaquaculture, an activity with high labour requirements. Poor households areless likely to be involved in aquaculture. We have already noted that thoseengaged in aquaculture are typically better off than average. Analysis of the datashows that the return, in terms of income earned per unit time spent, is higheron average in aquaculture compared to crop cultivation, though it is also riskierin that the return is also more variable (McCoy, McKay, and Perge 2010).Households that speak Vietnamese are slightly less likely to engage in

aquaculture, perhaps reflecting the geographic pattern of this activity—withthis activity, for instance, being relatively popular in the Northern Uplands.

Table 4.8 Continued

Sales of rice Cultivation of cash crops



RE probitmodel



RE probitmodel

coef/t coef/t coef/t coef/t coef/t coef/t

(0.569) (1.389) (0.699) (�1.635) (�2.106) (�1.791)If can read and write 0.002 0.003 0.078 �0.016 �0.016 �0.291

(0.080) (0.099) (1.023) (�1.640) (�1.510) (�1.357)2006 dummy (dropped) (dropped)2008 dummy 0.073*** �0.011

(4.147) (�1.372)2010 dummy 0.076*** �0.010

(4.247) (�1.189)2012 dummy 0.057*** �0.040***

(3.127) (�4.730)2014 dummy 0.106*** �0.002

(4.857) (�0.198)_cons 0.275** 0.630*** 0.178 �0.080 0.216*** �7.355***

(2.088) (5.545) (0.726) (�1.286) (4.997) (�8.529)Household FEs y yDistrict FE y yProvince FEs y y

/lnsig2u �0.563*** 2.155***(�6.500) (22.406)

Number of observations 7,335 7,335 7,335 8,004 8,004 8,004Rho 0.432 0.363 0.818 0.896Sigma_u 0.356 0.755 0.344 2.937Sigma_e 0.408 0.162

Note: *** p < 0.01, ** p < 0.05, * p < 0.1. OLS¼Ordinary Least Squares; RE¼ random effects; FE¼fixed effects.

Source: Authors’ calculation based on VARHS 2006–14 panel dataset.


84

Table 4.9 Regression results for correlates of engagement in aquaculture

Base model With shocks Shocks andinvestments

Pooled OLSwith district FE

Household FEpanel model

RE probitmodel

RE probitmodel

RE probitmodel

coef/t coef/t coef/t coef/t coef/t

Percentage ofland area withrestrictions

�0.0171** �0.00631 �0.0665 �0.0720 0.0306

(0.00856) (0.00773) (0.0673) (0.0677) (0.0818)Total land area 6.69e-07*** 2.44e-07 2.41e-06** 2.33e-06* 4.72e-06*

(2.01e-07) (2.27e-07) (1.22e-06) (1.22e-06) (2.43e-06)Percentage of

land areairrigated

0.00847 0.00110 0.0246 0.0274 0.111

(0.0116) (0.00913) (0.0793) (0.0796) (0.0907)If household

received credit0.0272*** 0.0443*** 0.390*** 0.396*** 0.269***

(0.00724) (0.00701) (0.0613) (0.0617) (0.0705)If household has

red book for itsland

0.0107 �0.00643 0.0523 0.0416 0.135

(0.0121) (0.0130) (0.115) (0.115) (0.131)Distance to

nearest road�5.36e-06 3.97e-07 �0.000148 �0.000141 �0.000161

(4.46e-05) (3.96e-05) (0.000810) (0.000764) (0.000853)If market in

commune0.000191 �0.0368*** �0.393*** �0.406*** �0.258***

(0.0124) (0.00783) (0.0773) (0.0779) (0.0828)If household does

wage work�0.0101 0.00696 �0.00705 �0.00881 0.0345

(0.00749) (0.00793) (0.0653) (0.0655) (0.0763)Household size 0.00693*** 0.0116*** 0.0632*** 0.0651*** 0.0188

(0.00231) (0.00331) (0.0231) (0.0232) (0.0255)If member of

farmers’ union0.00384 �0.00264 0.0106 0.00248 �0.0198

(0.00750) (0.00759) (0.0649) (0.0653) (0.0755)If member of

women’s union0.0132* �0.00623 0.0545 0.0567 0.0144

(0.00777) (0.00800) (0.0679) (0.0684) (0.0767)If owns a radio 0.00254 �0.00284 �0.00162 �0.000920 0.0129

(0.00951) (0.00890) (0.0770) (0.0776) (0.0901)If has own

transport0.0107 0.00504 0.132 0.116 0.0818

(0.0113) (0.00999) (0.0908) (0.0913) (0.102)If of Kinh ethnicity 0.0245 0.0516 0.0448 0.0304 �0.00448

(0.0172) (0.0397) (0.147) (0.148) (0.146)If female headed 0.0321*** 0.0338* 0.414*** 0.406*** 0.295**

(0.00947) (0.0201) (0.125) (0.125) (0.124)If speaks

Vietnamese�0.0451 �0.0727** �0.453* �0.447* �0.160

(0.0288) (0.0308) (0.234) (0.234) (0.341)If poor (MOLISA) �0.0613*** �0.0253** �0.415*** �0.421*** �0.369***

(0.00985) (0.0106) (0.0944) (0.0947) (0.112)

(continued )


85

Table 4.9 Continued

Base model With shocks Shocks andinvestments

Pooled OLSwith district FE

Household FEpanel model

RE probitmodel

RE probitmodel

RE probitmodel

coef/t coef/t coef/t coef/t coef/t

Age 0.000724** 0.000574 0.00212 0.00212 0.00358(0.000328) (0.000451) (0.00301) (0.00301) (0.00323)

If can read andwrite

0.0394*** �0.00621 0.239* 0.232* 0.215

(0.0134) (0.0158) (0.129) (0.129) (0.144)If faced a naturalshock

�0.139* �0.0357

(0.0760) (0.0838)If faced aneconomic shock

0.0723 0.0714

(0.133) (0.139)If made cash investmentsin aquaculture land

2.227***

(0.122)2006 dummy 0.0592***

(0.0152)2008 dummy 0.0308**

(0.0157)2010 dummy 0.0366**

(0.0158)2012 dummy �0.00412

(0.0134)_cons �0.165* 0.0597 �3.678*** �3.620*** �3.545***

(0.0882) (0.0581) (0.408) (0.409) (0.492)

Household FEs YDistrict FE YProvince FEs Y Y Y

/lnsig2u 0.639*** 0.642*** 0.216*(0.0972) (0.0976) (0.128)

Observations 8,466 8,466 8,466 8,400 6,568R-squared 0.204 0.018Number of unikid 1,959 1,959 1,959 1,897Rho 0.512 0.655 0.655 0.554Sigma_u 0.255 1.376 1.378 1.114Sigma_e 0.249

Note: *** p < 0.01, ** p < 0.05, * p < 0.1. RE¼ random effects; FE¼fixed effects.

Source: Authors’ calculation based on VARHS 2006–14 panel dataset.


86

The results in the fourth and fifth columns of Table 4.9 add further variables tothe base specification. In the fourth column, natural shocks reduce the likeli-hood of having been engaged in aquaculture, given that they make thisactivity infeasible in a particular year, whereas past investment in aquacultureis, of course, positively associated with undertaking the activity. This is anactivity requiring a significant degree of planning and investment. The mainresults here are relatively consistent across the different model specifications,including in relation to the variables added in the fourth and fifth columns.

These regression results are initial estimates, focusing only on contempor-aneous correlations, and can only identify associations. They do, though,confirm several of the patterns already suggested in the descriptive analysisin Sections 4.4 and 4.5. In the case of rice, those engaged in selling aregenerally those cultivating on a larger scale. Geographic factors are importantin relation to both cash crops and aquaculture, which can reflect many factorsincluding climatic conditions as well as, potentially, local policies. In general,there is a clear association between engagement in these commercial activitiesand being better off. But, of course, it is not possible to say anything aboutcausality based on this; better-off households may be better placed to beengaged in commercial activity (e.g. by having more land), but householdsmay also become better off by being engaged in these activities. In reality,both processes are probably at work.

4.7 Conclusions

This chapter has presented an initial analysis of the extent of commercializa-tion of agriculture in these twelve provinces of rural Viet Nam, focusing on thefive waves of the VARHS panel. What is clear, first, is the continuing import-ance of agriculture in rural Viet Nam, and this remains true for householdswhomay now earnmore of their income fromwages or other sources. Second,agriculture is increasingly commercialized in rural Viet Nam, with rice salesbeing the main area of commercialization. The vast majority of rural house-holds grow rice, of whom around half sell in any given year. There arevariations in this by geography and wealth, but, unsurprisingly, those produ-cing more and using more inputs are more likely to sell. However, the panelshows that not many households sell consistently from one year to another.Presumably the decision to sell reflects the scale of production in a given yearand perhaps available opportunities. The exception to this is Long An, wherethe activity is on a much larger scale, with households selling more, muchmore regularly, compared to any other provinces.

Cash crop production and aquaculture are clearly also commercial activitiesundertaken by a non-negligible minority of these surveyed households,


87

although strong geographic patterns exist, in part reflecting the suitability ofdifferent locations for these activities. Unsurprisingly, given its nature, usuallyinvolving tree crops, cash crop activity shows substantial persistence over timein the panel, but in aquaculture there are many fluctuations from one year toanother. While this is potentially a high-return activity for households, it isalso relatively labour-intensive and relatively risky. It may therefore be harderfor households to guarantee a worthwhile return from this activity every year.One thing that clearly emerges from this initial analysis of the data is a

strong association between commercialization and wealth, and a two-wayprocess of causality is probably at work here. But it is almost certainly the casethat increased commercialization of agricultural activities in rural Viet Nam hasbeen an important contributor to the impressive rural poverty reduction thecountry has experienced.There ismuch scope to analyse these questions inmore depth in subsequent

work, in particular exploiting further the panel features of the dataset. This isexpected to allow clearer conclusions to be drawn about the nature of thefactors facilitating commercialization in agriculture in rural Viet Nam, includ-ing the ability to engage consistently in commercial activities over time. Theseissues will be addressed in more detail in future work.

Acknowledgements

We are very grateful for helpful comments received when presenting a draft of thisstudy to the Central Institute of Economic Management (CIEM) in Hanoi; as well as forcomments received based on earlier presentations of this work to CIEM. We are alsovery grateful for advice and suggestions from Finn Tarp and others in the VARHSanalysis team throughout this work.

References

Abbott, P., F. Tarp, and C. Wu (forthcoming). ‘Structural Transformation, BiasedTechnological Change and Employment in Vietnam’, European Journal of DevelopmentResearch. DOI: 10.1057/ejdr.2015.64.

Aksoy, A. and A. Isik-Dikmelik (2007). The Role of Services in Rural Income: The Case ofViet Nam. World Bank Policy ResearchWorking Paper 4180. Washington, DC:WorldBank.

Alene, A. D., V. M. Manyong, G. Omanya, H. D. Mignouna, M. Bokanga, andG. Odhiambo (2008). ‘Smallholder Market Participation under Transaction Costs:Maize Supply and Fertilizer Demand in Kenya’. Food Policy, 33: 318–28.

Barrett, C. B. (2008). ‘Smallholder Market Participation: Concepts and Evidence fromEastern and Southern Africa’. Food Policy, 33(4): 299–317.


88

Benjamin, D. and L. Brandt (2004). ‘Agriculture and Income Distribution in RuralVietnam under Economic Reforms: A Tale of Two Regions’. In P. Glewwe,N. Agrawal, and D. Dollar (eds), Economic Growth, Poverty and Household Welfare inVietnam. Washington DC: World Bank.

Benjamin, D., L. Brandt, B. Coelli, B. McCaig, L.-H. Nguyen, and T. Nguyen (2009).‘Crop Output in Vietnam, 1992 to 2006: An Analysis of the Patterns and Sources ofGrowth’. Report prepared for the World Bank.

Boughton, D., D. Mather, C. B. Barrett, R. Benfica, D. Abdula, D. Tschirley, andB. Cunguara (2007). ‘Market Participation by Rural Households in a Low-IncomeCountry: An Asset-Based Approach Applied toMozambique.’ Faith and Economics, 50:64–101.

Cadot, O., L. Dutoit, and M. Olarreaga (2006). How Costly is it for Poor Farmers to LiftThemselves Out of Subsistence? World Bank Policy Research Working Paper 3881.Washington DC: World Bank.

Cazzuffi, C. and A. McKay (2012). ‘Rice Market Participation and Channels of Sale inRural Viet Nam’. Paper presented at International Association of Agricultural Econo-mists (IAAE) Triennial Conference, Foz do Iguaçu, Brazil, August.

Cazzuffi, C., A. McKay, K. D. Luu, T. L. Nguyen, and D. M. Thuy (2011). ‘Constraintsto Market Participation in Agriculture in Viet Nam’. CIEM Working Paper. Hanoi:CIEM.

Glewwe, P. (2004). ‘An Overview of Economic Growth and Household Welfare inVietnam in the 1990s’. In P. Glewwe, N. Agrawal, and D .Dollar (eds), EconomicGrowth, Poverty and Household Welfare in Vietnam. Washington DC: World Bank.

Goetz, S. J. (1992). ‘A SelectivityModel of Household FoodMarketing Behaviour in Sub-Saharan Africa’. American Journal of Agricultural Economics, 74: 444–52.

Heltberg, R. andF.Tarp (2002). ‘Agricultural SupplyResponse andPoverty inMozambique’.Food Policy, 27(1): 103–24.

Key, N., E. Sadoulet, and A. de Janvry (2000). ‘Transaction Costs and Agricultural House-hold Supply Response’. American Journal of Agricultural Economics, 82(1): 245–59.

Levinsohn, J. A. and M. McMillan (2007). ‘Does Food Aid Harm the Poor: HouseholdEvidence from Ethiopia’. In A. Harrison (ed.), Globalization and Poverty. Chicago:University of Chicago Press.

McCaig, B. andN. Pavcnik (2013).Moving out of Agriculture: Structural Change in Vietnam.NBER Working Paper No. 19616. Cambridge, MA: National Bureau of EconomicResearch.

McCoy, S., A.McKay, and E. Perge (2010). ‘An Analysis of Small Scale Fisheries ActivitiesinVietNam: Aquaculture and InlandCapture’. Report toCIEMandDanida, December.Hanoi: CIEM.

McKay, A. and F. Tarp (2015). ‘Distributional Impacts of the 2008 Global Food PriceSpike in Vietnam’. In D. E. Sahn (ed.), The Fight against Hunger and Malnutrition: TheRole of Food, Agriculture and Targeted Policies. Oxford: Oxford University Press.

McKay, A., C. Cazzuffi, and E. Perge (2015). Commercialization in Agriculture in Rural VietNam, 2006–14. WIDER Working Paper 2015/096. Helsinki: UNU-WIDER.

Markussen, T., F. Tarp, and K. van den Broeck (2011). ‘The Forgotten Property Rights:Evidence on Land Use Rights in Vietnam’. World Development, 39(5): 839–50.


89

Marsh, S. P., T. G. MacAulay, and P. Van Hung (2006). ‘Agricultural Development andLand Policy in Vietnam: AnOverview and Theoretical Perspective’. In T. G.MacAulay,S. P. Marsh, and P. Van Hung, Agricultural Development and Land Policy in Vietnam.Australian Centre for International Agricultural Research, Canberra, Australia.

Nyoro, J., M. W. Kiiru, and T. S. Jayne (1999). Evolution of Kenya’s Maize MarketingSystems in the Post-Liberalisation Era. Tegemeo Institute of Agricultural Policy andDevelopment Working Paper 2A. Nairobi: Tegemeo Institute of Agricultural Policyand Development.

Ouma, E., J. Jagwe, G. A. Obare, and S. Abele (2010). ‘Determinants of SmallholderFarmers’ Participation in Banana Markets in Central Africa: The Role of TransactionCosts’. Agricultural Economics, 42(2): 111–22.

Rios, A., G. Shively, and W. Masters (2009). ‘Agricultural Productivity and HouseholdMarket Participation: Evidence from LSMS Data’. Paper presented at the 27th Inter-national Conference of Agricultural Economists. Beijing, August.

Stephens, E. C. and C. B. Barrett (2006). ‘Incomplete Credit Markets and CommodityMarketing Behaviour’. Cornell University Working Paper. Ithaca, NY: CornellUniversity.


90

5

The Rural Non-Farm Economy

Christina Kinghan and Carol Newman

5.1 Introduction

The diversification of economic activity away from the agricultural sector is akey characteristic of economic development. As households transition intodifferent economic activities, it is likely that some will disproportionatelybenefit while others will be left behind (see Chapter 10). Understanding theoutcomes from diversification in addition to the determinants that prompthouseholds to diversify is therefore of great importance to policy makers.Given Vietnam’s emphasis on promoting equity and social inclusion (WorldBank 2016), establishing the extent to which all households have the sameopportunities to pursue diversification is of particular importance.

This chapter examines the extent to which rural Vietnamese householdshave, in recent times, diversified away from own-farm agriculture into wagedemployment and entrepreneurial activities, and the impact of this diversifica-tion on welfare outcomes. We also examine whether the welfare outcomes forhouseholds that participate in more than one type of economic activity aresuperior to those who remain specialized in agriculture, and the characteristicsof households that are most likely to diversify.

The impact of diversification into non-farm activities on rural householdshas been well documented. Overall, the literature in this area concludesthat, while diversification is positively correlated with income and wealth(Economica Vietnam 2013), it also has the potential to increase inequality,as households with favourable initial characteristics and conditions may dis-proportionately benefit. This highlights the potential for a dichotomous out-come from non-farm activity, where poorer households partake in low-returnactivities and wealthier households undertake high-return activities. Differingoutcomes from participating in non-farm activities can also be observed when

diversification is as a result of ‘push’ factors such as shocks, risk reduction, andsurvival. These broad conclusions motivate the analysis of diversificationundertaken in this chapter.Imai, Gaiha, and Thapa (2015) observe significantly higher per capita con-

sumption, as a proxy for poverty reduction, for households participating in thenon-farm sector in both Viet Nam and India. Access to non-farm work alsodecreased vulnerability to shocks, reducing risk. However, effects were signifi-cantly higher for households participating in skilled employment compared tothose working in unskilled/manual positions. Hoang, Pham, and Ulubasoglu(2014), suggest that diversification can act as a strong tool for poverty allevi-ation in Viet Nam. They find that an additional member of the household,working in a non-farm activity, decreases the probability of poverty by 7–12per cent and can increase household expenditure by up to 14 per cent over atwo-year period. Furthermore, their results indicate that a reduction in hoursworked on the farm due to non-farm work does not lead to a reduction inincome earned from agricultural activities. Bezu, Barrett, and Holden (2012)also find a strong positive relationship between a household’s non-farmincome share and its subsequent expenditure growth, for both poor and well-off households in Ethiopia. Yet, relatively wealthier households benefitedmore from off-farm activity than poorer households did.Similarly, Lanjouw, Murgai, and Stern (2013) found that non-farm diversi-

fication in India not only led to increased incomes and reductions in poverty,but that it was also instrumental in breaking down barriers to economicmobility among the poorest segments of society. Coupled with diversification,however, they highlighted rising income inequality at village level and thepotential impact this inequality may have on social cohesiveness. Birthaland colleagues (2014) stated that poorer households tended to diversify intolow-return activities and that this diversification had an unequalizing effecton the income distribution, but a positive impact on household income forrural households in India. A report undertaken by the Development AnalysisNetwork (2003), found that while non-farm employment was important forjob creation in Viet Nam, it significantly widened the non-farm income gapbetween rich and poor, hence contributing to social inequality. This researchemphasized both the opportunities to be gained from diversification and alsothe potential for growing income inequality among rural households.Regarding the determinants of diversification into non-farm activity,

Olugbire and colleagues (2012) consider the household characteristics associ-ated with participation in non-farm employment and entrepreneurship inNigeria. They conclude that education, gender, land, and household size arekey determinants of participation in non-farm waged employment, whereasvalue of assets, access to credit, social capital, household, and land sizeare important determinants of non-farm entrepreneurship. Similarly, Ackah


92

(2013) finds land size, education past primary level, and gender are importantdeterminants of diversification in Ghana, with females more likely to beengaged in non-farm work. Education past secondary school is of particularimportance for stable waged employment. Benedikter and colleagues (2013)also note a correlation between enterprise size and owner education. Theyfind that the level of savings, prior work experience, and family relations/inheritance were key factors in establishing a non-farm enterprise in theMekong Delta in Viet Nam. Micevska (2008) emphasizes the importance ofeducation for diversification, finding that individuals with higher educationlevels tend to diversify into high-return non-farm activities, with low-returnactivities pursued by those with limited education levels. This in turn influencesthe level of income generated by diversification. Overall, this indicates thatresource-poor, less educated households may face significant barriers to entryinto non-farm activity.

Giesbert and Schindler (2012) examine welfare dynamics among ruralhouseholds in Mozambique. They find that drought has a negative impacton a household’s asset accumulation, but households in which at least onemember has regular non-farm work experience less adverse asset growth froma drought than those without non-farm wage opportunities—suggesting thatincome diversification has a positive impact in the aftermath of an exogenousshock. Looking at the impact of shocks on diversification in Ethiopia, Porter(2012) finds that households that increase non-crop income as a result ofrainfall shocks can effectively cancel out the negative impact on crop income.Bezu and Barrett (2012) also conclude that shocks reducing agricultural incomecan trigger transition into high-return non-farm activities, with shocks towealth resulting in transition into low-return non-farm activities.

At a broader level, Haggblade, Hazell, and Reardon (2010) highlight theimportance of agricultural development in determining whether diversifica-tion will be primarily as a result of ‘pull’ factors into high-return activitiesor ‘push’ factors into low-return activities. They posit that this is the resultof linkages between agriculture and diversification. Positive linkages includerising incomes stimulating demand for products and services, increased prod-uctivity freeing up labour for non-farm work, and demand for seeds andfertilizers, all of which stimulate a productive non-farm sector. In contrast,where the agricultural sector is stagnant or declining, yet population growthis increasing, linkages such as low labour productivity, rising landlessness,and limited household purchasing power will induce diversification intolow-return activities. Vijverberg and Haughton (2002) also emphasized therole of transition in determining the extent and type of diversification. Theyposit that household enterprises rise in importance as an economy developsbut are then replaced by better economic opportunities. In this way, anenterprise can play an important bridging role, providing an alternative to


93

agriculture when waged employment is scarce, but with less appeal thanexternal employment.In summary, the literature suggests that diversification into non-farm activ-

ities by rural households has a positive impact on overall household incomes/expenditures. However, the impact of diversification on the income distribu-tion and ensuing inequalities between households is less clear. Differingreturns are evident for households based on their individual characteristics,which may determine whether they diversify into high- or low-return activ-ities. Whether diversification is in response to a shock, and hence promptedby ‘push’ factors, or due to favourable endowments possessed by a household,can lead to heterogeneous welfare outcomes from non-farm activity. This is ofimportance to policy makers; facilitating and indeed encouraging diversifica-tion of household incomes should result in improved welfare outcomes, yetthis may be at the cost of rising inequalities and divisions in society.The rest of the chapter proceeds as follows. Section 5.2 describes the pattern

of diversification of rural households in Viet Nam while Section 5.3 docu-ments the transition from specialized agriculture into other activities.Section 5.4 presents an empirical analysis of the impact of diversification onwelfare and of the factors that determine the transition from agriculture.Section 5.5 concludes with a discussion of the key findings and recommenda-tions for both policy and future research in this area.

5.2 Description of Non-Farm Activities of Households

Table 5.1 details the economic activities that the sample of Viet Nam Access toResources Household Survey (VARHS) households were involved in during the2008–14 time period. A household’s economic activity can fall into one ofeight categories based on engagement in agriculture, labour, enterprises, or acombination of these, and a category for households that are inactive. Theshare of households engaged only in agriculture has fallen steadily from2008–14, highlighting the micro-level structural transformation taking placein Viet Nam at the household level. Few households diversify away fromagriculture completely (consistent with the findings of the commune-levelanalysis presented in Chapter 3), but we observe a steady, albeit small, increasein the number of households specializing in labour or enterprises. The mostcommon form of diversification is supplementing agriculture with labour,which increases consistently throughout the sample period.Table 5.2 contains information on the proportion of income households

earn from agriculture, labour, enterprise, or other income sources (such as rentand transfers). This highlights the decreasing proportion of household incomeoriginating from agriculture and large increases in the importance of waged


94

employment. We also observe a decrease in income earned by enterprises in2012. This drop is potentially the result of poor macroeconomic conditions.

Looking at the characteristics of household enterprises, in Table 5.3 weobserve that more than half are led by a female household member. As only20 per cent of households have a female household head, it appears thatdiversifying by operating a non-farm enterprise is commonly undertaken byfemale household members to generate additional income for the family. Thisis in line with previous empirical research into non-farm household enter-prises in Viet Nam, which finds that the sector is becoming increasinglyfeminized (Oostendorp, Trung, and Tung 2009). Nearly 80 per cent of enter-prises do not have a business licence and so operate in the informal sector ofthe economy, with little evidence of increasing formalization of enterpriseactivities over the years of the survey.

Over 80 per cent of household enterprises are operated by one to threeindividuals, with a further 10 per cent having four to six workers. Only 5 percent have more than seven people working in the enterprise. We can also lookat howmany of these individuals receive a wage for their work. Approximately85 per cent of enterprises do not pay a wage to those working in the enterprise,10 per cent of enterprises pay a wage to between one and three employees, andonly 5 per cent of enterprises pay a salary to more than four workers. Thesedescriptive statistics are in line with the findings that diversification into

Table 5.1 Economic activities of households, 2008–14

Per centHH

Ag.only

Labouronly

Ent. only Ag. &labour

Ag. &ent.

Ag., labour,& ent.

Labour& ent.

Noactivity

2008 25.16 4.09 2.39 40.62 11.41 11.50 2.44 2.392010 22.38 4.45 3.03 41.91 12.10 10.04 2.93 3.162012 20.59 5.73 3.58 43.15 9.35 10.45 2.43 4.722014 19.53 5.64 3.76 45.62 6.79 10.36 3.39 4.91

Note: n = 2,181.

Source: Authors’ calculation based on survey data comprised from VARHS for the years 2008 to 2014.

Table 5.2 Proportion of income earned from different economic activities,2008–14

Per cent HH Agriculture Labour Enterprise Other

2008 34.76 28.15 12.63 24.362010 23.36 31.26 13.67 31.662012 23.00 32.92 3.85 40.112014 23.80 44.35 12.28 19.54

Note: n = 2,181.



95

non-farm activity is more likely to be undertaken by low-income householdsand often in response to a shock. While welfare enhancing, the vast majorityof enterprises tend to be operated informally and on a low scale, as a basicmeans for households to generate additional income. However, almost allhouseholds were required to invest in the enterprise in order to start doingbusiness, with more than 90 per cent of households stating that an initialinvestment was needed to diversify into this activity.The age and education of enterprise managers are also important when

examining the key characteristics of these household enterprises. The averageage of an enterprise manager is 45, with a wide disparity in ages, ranging from11–91 years old. On average, enterprise managers have completed eight yearsof schooling. Finally, the most popular industries were processing and manu-facturing (30 per cent), wholesale and retail trade (28 per cent), and accom-modation and food services (9 per cent). A full list of the industry sectors isgiven in Table 5.A1 of the Appendix.Regarding external employment, Table 5.4 shows an increase both in house-

holds with a member working externally and the number of individualsworking externally, over the years of the survey. The number of householdsthat do not have any kind of external employment fell from 41 per centin 2008 to 35 per cent in 2014, and the number of households withthree household members working externally increased from 7 per cent to

Table 5.3 Enterprise characteristics

2008 2010 2012 2014 Total

Gender ManagerFemale 328 (55%) 331 (54%) 291 (52%) 254 (48%) 1,204 (52%)Male 271 (45%) 282 (46%) 272 (48%) 276 (52%) 1,101 (47%)Formal EnterpriseInformal 470 (78%) 471 (77%) 444 (79%) 409 (77%) 1,794 (78%)Formal 129 (22%) 142 (23%) 119 (21%) 121 (23%) 511 (22%)Total Labour1–3 workers 508 (86%) 509 (84%) 469 (84%) 428 (81%) 1,914 (84%)4–6 workers 61 (10%) 71 (12%) 58 (10%) 72 (14%) 262 (11%)7–62 workers 25 (4%) 28 (5%) 31 (6%) 29 (5%) 113 (5%)Total Paid Labour0 employees 526 (88%) 528 (86%) 484 (86%) 425 (80%) 1,963 (85%)1–3 employees 48 (8%) 55 (9%) 52 (9%) 72 (14%) 227 (10%)4–60 employees 25 (4%) 30 (5%) 27 (5%) 33 (6%) 115 (5%)Needed InvestmentNo 51 (9%) 28 (5%) 25 (4%) 20 (4%) 124 (5%)Yes 548 (91%) 585 (95%) 538 (96%) 510 (96%) 2,181 (95%)Variable Observations Mean Std. Dev Min. MaxAge 2,297 44.98 11.90 11 91Education 2,297 7.58 3.50 0 12



96

9 per cent. However, while a large number of households have membersworking externally, less than half of these households have at least onemember with a formal labour contract. This indicates that the kind of employ-ment undertaken by diversifying households may be informal.

We see further evidence of this when we examine the employers of house-hold members. Approximately 70 per cent of the households engaged inexternal employment state that this employment is with another individualor household, compared to 25 per cent with a member employed by a gov-ernment or state enterprise, and less than 20 per cent employed by a privateVietnamese firm. In terms of the location for these activities, employment iswidely dispersed. Around 60 per cent of households have a householdmemberworking within the commune, 25 per cent working in another communewithin the district, and 30 per cent working outside the district. Finally, themost popular sectors for external employment are construction and engineering

Table 5.4 External employment descriptive statistics

2008 2010 2012 2014 Total

Members working0 900 (41%) 894 (41%) 833 (38%) 761 (35%) 3,388 (39%)1 604 (28%) 620 (29%) 648 (30%) 594 (27%) 2,466 (28%)2 458 (21%) 417 (19%) 475 (22%) 546 (25%) 1,896 (22%)3 142 (7%) 161 (7%) 153 (7%) 184 (9%) 640 (7%)4–10 70 (3%) 85 (4%) 68 (3%) 91 (4%) 314 (4%)Labour contractNo 793 (62%) 830 (65%) 813 (60%) 837 (59%) 3,273 (62%)Yes 481 (38%) 453 (35%) 531 (40%) 578 (41%) 2,043 (38%)Member employed by:Individual/householdNo 363 (28%) 366 (29%) 436 (32%) 451 (32%) 1,616 (30%)Yes 911 (72%) 917 (71%) 908 (68%) 964 (68%) 3,700 (70%)Government/state enterpriseNo 974 (76%) 980 (76%) 1,016 (76%) 1,069 (76%) 4,039 (76%)Yes 300 (24%) 303 (24%) 328 (24%) 346 (24%) 1,277 (24%)Vietnamese private firmNo 1,098 (86%) 1,111 (87%) 1,081 (80%) 1,094 (77%) 4,384 (82%)Yes 176 (14%) 172 (13%) 263 (20%) 321 (23%) 932 (18%)Location employment:Within communeNo 524 (41%) 489 (38%) 466 (35%) 443 (31%) 1,922 (36%)Yes 757 (59%) 798 (62%) 882 (65%) 977 (69%) 3,414 (64%)Another commune in districtNo 982 (77%) 938 (73%) 955 (71%) 1,048 (74%) 3,923 (74%)Yes 292 (23%) 345 (27%) 389 (29%) 367 (26%) 1,393 (26%)Outside districtNo 786 (62%) 910 (71%) 1,024 (76%) 1,016 (72%) 3,736 (70%)Yes 488 (38%) 373 (29%) 320 (24%) 399 (28%) 1,580 (30%)



97

(24 per cent), processing and manufacturing (19 per cent), and agriculture,forestry, and aquaculture (17 per cent). A full list of industry sectors is given inTable 5.A2 in the Appendix to this chapter.

5.3 Diversification and the Transition from Agriculturein Viet Nam

Table 5.5 presents detailed transition matrices for households, demonstrat-ing the extent of movement between different types of economic activitiesover time.A strong pattern of movement away from specializing in agriculture is

evident. Almost 50 per cent of households involved only in agriculture in2008 had diversified into another economic activity in 2010. Of these house-holds, 25 per cent combined agriculture with labour and 10 per cent com-bined agriculture with a non-farm enterprise. There is also evidence of furtherdiversification by those households involved in agriculture and labour. While67 per cent remained in this category, approximately 8 per cent diversifiedfurther by establishing a household enterprise. Thirteen per cent of house-holds engaged with agriculture and enterprises further diversified into paidemployment. We do observe some reversion to agriculture only for house-holds that combined agriculture with labour or enterprises (14 per cent and11 per cent, respectively). However, this likely reflects job losses and enter-prise failure.This pattern is consistent in the 2010–12 and 2012–14 time periods, with

further movement away from agriculture specialism. In both years, more than50 per cent of those who were previously engaged in agriculture only diversi-fied their economic activities. Interestingly, those households that wereinvolved in enterprise only show a strong tendency to move towards labouronly or labour and enterprise, especially in the 2010–12 and 2012–14 periods.This may reflect the uncertainty associated with operating an enterprise,compared to the stability of waged employment. Approximately 12 per centof households with enterprise only transitioned to labour only in 2010–12and 2012–14. Thirteen per cent supplemented enterprise operation withlabour in 2010–12 and this rose to almost 18 per cent in 2012–14. This mayalso be reflective of a more tumultuous operating environment in the 2012–14period, due to the global recession. This potentially impacted on the viabilityof sustaining household incomes through enterprise activity alone.The transition matrices highlight the large variation and movement

between the economic activities of these households. It is evident that house-holds rely on a variety of sources to generate income. In particular, we observeamovement away from specialization with agriculture as the solitary source of


98

Table 5.5 Economic activity transition matrices, 2008–14

2008–10 Ag. only Labour only Ent. only Ag. & labour Ag. & ent. Ag., labour, & ent. Labour & ent. No activity

Ag. only 52.83 0.37 0.00 25.41 10.05 7.68 0.00 3.66Labour only 3.37 48.31 4.49 21.35 1.12 2.25 13.48 5.62Ent. only 0.00 0.00 53.85 5.77 13.46 5.77 13.46 7.69Ag. & labour 14.49 3.26 0.11 67.08 5.51 7.75 0.67 1.12Ag. & ent. 11.29 1.21 6.85 20.16 45.97 12.50 0.81 1.21Ag., labour, & ent. 8.40 3.20 2.00 39.60 13.20 28.40 4.40 0.80Labour & ent. 0.00 16.98 15.09 7.55 9.43 0.00 47.17 3.77No activity 34.62 5.77 5.77 5.77 0.00 1.92 1.92 44.23

2010–12 Ag. only Labour only Ent. only Ag. & labour Ag. & ent. Ag., labour, & ent. Labour & ent. No activityAg. only 48.16 2.46 0.41 31.76 6.56 3.28 0.20 7.17Labour only 4.12 52.58 2.06 27.84 3.09 2.06 4.12 4.12Ent. only 3.03 12.12 54.55 1.52 6.06 6.06 12.12 4.55Ag. & labour 13.35 3.61 0.98 68.38 4.92 7.00 0.33 1.42Ag. & ent. 18.18 0.00 4.17 18.94 35.23 21.21 0.76 1.52Ag., labour, & ent. 10.96 2.28 0.91 35.16 11.42 36.53 1.83 0.91Labour & ent. 0.00 15.63 20.31 4.69 1.56 6.25 48.44 3.13No activity 20.29 8.70 4.35 4.35 1.45 2.90 0.00 57.97

2012–14 Ag. only Labour only Ent. only Ag. & Labour Ag. & ent Ag., labour, & ent. Labour & ent. No activityAg. only 48.33 1.34 0.89 33.85 5.12 4.45 0.22 5.79Labour only 5.60 48.00 4.80 20.00 0.00 6.40 8.80 6.40Ent. only 0.00 11.54 47.44 5.13 5.13 5.13 17.95 7.69Ag. & labour 14.35 2.87 0.21 71.94 2.44 6.70 0.64 0.85Ag. & ent. 14.22 0.49 6.86 23.04 30.39 23.04 1.47 0.49Ag., labour, & ent. 6.14 0.44 2.63 37.72 14.47 32.89 4.82 0.88Labour & ent. 0.00 5.66 20.75 1.89 5.66 15.09 50.94 0.00No activity 23.30 15.53 1.94 2.91 0.00 0.97 0.97 54.37

Note: n = 2,181.


income. This chapter aims to explore whether this transition leads to improve-ments in household welfare. To do this, we utilize two different welfareindicators: food expenditure/consumption and household income.Expenditure on food is the key welfare indicator used in our analysis. It is

less likely to suffer from measurement error than household income and istherefore a more reliable and accurate measure of the welfare benefits fromdiversification (Meyer and Sullivan 2011). The variable is constructed byaggregating the value of a set of food items consumed by the household in

Table 5.6 Welfare measures, 2008–14

Real incomeper capita

Real food expenditureper capita

2008Ag. only 956 237Labour only 1,752 382Ent. only 2,769 491Ag. & labour 1,117 298Ag. & ent, 2,027 397Ag., labour, & ent. 1,503 341Labour & ent. 1,745 415Total 1,318 311

2010Ag. only 1,223 312Labour only 2,139 377Ent. only 3,400 504Ag. & labour 1,369 321Ag. & ent. 2,173 388Ag., labour, & ent. 2,014 400Labour & ent. 2,531 419Total 1,649 349



Note: n = 2,181.

Source: Authors’ calculation based on survey data comprised from VARHS for the years2008 to 2014.


100

the previous month and is converted to real terms using a national food priceindex. We also consider total household income in real 2014 values.

Table 5.6 contains the group means for these two welfare measures byeconomic activity undertaken by the household. Focusing on the total figuresfirst, we see increases in each time period in real household income (percapita) and real food expenditure (per capita). Disaggregating by economicactivity, it is evident that average income is highest for households specializ-ing in enterprise activity only in each year. Households’ specializing in agri-culture, however, have the lowest income levels, below average for the groupas a whole. It appears on first glance, therefore, that any kind of diversificationleads to income improvements compared to remaining in agriculture only.Operating an enterprise is also positively correlated with high levels of foodexpenditure.

Average food expenditure (real per capita) is highest in each time period forhouseholds in either enterprise only or enterprise and labour categories.Households with labour only also have higher than average food expenditure,particularly in the later years. Again this highlights the welfare benefits ofmovement away from agriculture alone. Overall, these descriptive statisticshighlight the potential welfare-enhancing outcomes of non-farm diversifica-tion. However, these relationships will be formally examined in the ensuingempirical analysis in Section 5.4.

5.4 Empirical Analysis

In this section we explore further the impact of diversification on householdwelfare. Identifying a causal relationship between income diversification iscomplicated by the possibility that households may self-select into moreproductive activities. In other words, richer or wealthier households maychoose to diversify rather than diversification in itself leading to higher levelsof income or wealth. Any econometric model used to identify the effect ofdiversification on welfare must therefore control for all factors, observed orotherwise, that impact on both the welfare of the household and its decisionto diversify its income sources.

Using the balanced panel of data fromVARHS for the 2008–14 period allowsus to control for self-selection in two ways. First, with the inclusion of house-hold fixed effects, all time invariant characteristics of households are con-trolled for in the analysis, including the households’ initial wealth andincome levels. Second, the availability of lags allows past values of incomeand wealth to be controlled for in the analysis. As such, the impact of bothlong-term and transitory changes in income and wealth on welfare will becontrolled for, allowing us to isolate the specific impact of diversification.


101

We focus on consumption as our outcome measure of interest and control forhousehold fixed effects, past income, and wealth to address the self-selectionproblem. The model we estimate is as follows:

Cit ¼ β1Sit þ β2Xit þ β3Xit�1 þ β4Incomeit�1 þ β5Wealthit�1 þ αi þ τt þ eit

The key variables of interest are the sources of income of households. They areincluded in the vector Sit in the form of dummy variable indicators of thevarious categories described in Section 5.2, with households that are involvedin agriculture only (i.e. specialized agriculture) forming the base category. Thevector Xit includes household characteristics, namely household size, house-hold size squared, whether the household head is female, age of the house-hold head, age squared, the education level of the household head, thenumber of children in the household, whether the household is of Kinhethnicity, whether the head of household was born in the commune, andwhether the household is classified as poor by the authorities. Current periodwealth is also included as a control variable within this vector, measured usingan index of the assets held by an individual household. Further informationon how the asset index was constructed is detailed in Chapter 10. An add-itional complication with this specification is the need to control for currentperiod income of households, which is collinear with the sources of incomeand with the other control variables. If we assume that the generation of incomeis a dynamic process—in that past values will determine future values—the lagof income and the lag of other time-varying household characteristics(included in Xit�1) should serve as adequate controls. The model includeshousehold fixed effects, αi, and time dummies, τt ; eit is the statistical noiseterm. Summary statistics for each of the variables included in the analysis arepresented in Table 5.A3 of the Appendix.The results for the key variables of interest are presented in Table 5.7.1

Table 5.A4 of the Appendix details the full set of results for all of the explana-tory variables. The dependent variable is the log of real consumption percapita. Making the per capita adjustment is particularly important in thismodel given that diversification and food consumption will be related to thesize of the household. We also include household size to control for the factthat there may be economies of scale associated with food consumption inlarger households. A log transformation is used to reduce the impact ofoutliers and for ease of interpretation of the parameter estimates.Columns (1)–(3) reveal that households that are diversified are better off

than households that are specialized in agriculture. In particular, when allcontrol variables are included (column (3)), we find that households that are

1 We exclude households that report having no economic activities.


102

engaged in agriculture and with some other type of activity—waged employ-ment, a household enterprise, or both—have higher levels of consumptionper capita than those that are engaged in agricultural production only. Thecoefficient estimates suggest that, compared with households that are fullyspecialized in agricultural production, the fully diversified households do thebest, with consumption levels per capita that are 22 per cent higher thanhouseholds specialized in agriculture, followed by households that areengaged in agriculture and enterprise activities with consumption levels percapita that are almost 13 per cent higher, while households engaged inagriculture and waged employment have consumption levels per capita thatare almost 12 per cent higher.

Households with an enterprise are also better off in welfare terms thanhouseholds that are specialized in agriculture. Households that concentratesolely on household enterprise activities have consumption levels per capitathat are almost 20 per cent higher than those that are specialized in agriculture.Households with an enterprise and waged employment also have higher con-sumption levels but this difference is only marginally statistically significant.

In Table 5.8 we disaggregate the diversification of economic activities fur-ther, separating out households that moved out of specialized agriculturebetween survey rounds from other types of diversified households. We findthat the transition out of specialized agriculture is welfare enhancing. The percapita consumption of households that move from being engaged in

Table 5.7 Impact of diversification on household welfare

(1) (2) (3)

Ag. & labour 0.074*** 0.122*** 0.118***(0.027) (0.026) (0.027)

Ag. & ent. 0.125*** 0.149*** 0.127***(0.039) (0.038) (0.039)

Ag., labour, & ent. 0.163*** 0.229*** 0.224***(0.036) (0.037) (0.038)

Labour only 0.032 0.061 0.060(0.073) (0.073) (0.074)

Ent. only 0.170** 0.214*** 0.200***(0.072) (0.069) (0.071)

Labour & ent. 0.073 0.139** 0.135*(0.068) (0.067) (0.069)

HH characteristics No Yes YesLag controls No No YesTime dummies Yes Yes YesNumber of households 2,151 2,151 2,149Number of observations 6,263 6,238 6,150

Notes: Each specification includes household fixed effects. Standard errors clustered at the household level are presentedin parentheses. *** indicates significance at the 1% level, ** indicates significance at the 5% level, * indicates significanceat the 10% level.



103

agricultural production only into other types of production activities is almost14 per cent higher than those who remain specialized (column (1)). Whenthis is disaggregated by type of activity (column (2)) we find that this resultis driven by those households that diversify by entering into waged employ-ment or by both entering waged employment and adding an enterpriseactivity to their portfolio of production activities. Of the non-transitionhouseholds, those that are diversified also perform better, particularly thosethat are involved in both labour and enterprise activities. A full set of resultsincluding all of the explanatory variables is detailed in Table 5.A5 of theAppendix.We now turn our attention to exploring the characteristics of households

that transition out of agriculture. The dependent variable in this analysis takesa value of one if a household moved from specialized agricultural productionto some other combination of economic activities and zero otherwise. Asexplanatory factors, we include the full set of household characteristics, butat a lag so that we are considering the impact of past values of each character-istic on the decision to transition out of agriculture. A drawback of using ahousehold fixed effects approach in this case is that it factors out all timeinvariant household characteristics, observed and unobserved. It is, in fact,many of the time invariant characteristics, such as the ethnicity or gender ofthe household head, that are of most interest in determining what character-istics impact on the decision to diversify. As such, we estimate themodel using

Table 5.8 Impact of diversification out of agriculture on household welfare

(1) (2)

Transition out of ag. 0.138***(0.030)

Of which:Into ag. & labour 0.146***

(0.034)Into ag. & ent. 0.053

(0.070)Into ag., labour, & ent. 0.236***

(0.081)Into other 0.125*

(0.074)Control for activities of non-transition households Yes YesHH characteristics Yes YesLag controls Yes YesTime dummies Yes YesNumber of households 2,149 2,149Number of observations 6,150 6,150

Notes: Results for the control variables are presented in Table 5.A5 of the Appendix. Each specification includeshousehold fixed effects. Standard errors clustered at the household level are presented in parentheses. *** indicatessignificance at the 1% level, ** indicates significance at the 5% level, * indicates significance at the 10% level.



104

a random effects estimator, but control for time invariant characteristics byincluding the household specific means of the time-varying characteristics(the so-called Chamberlain–Mundlak adjustment).

The results are presented in Table 5.9. Column (1) reveals that higherincome households are less likely to transition out of agriculture. This suggeststhat diversification in the Vietnamese case is not driven by higher incomelevels. All types of income shocks (natural and economic) are positively relatedto the probability of transitioning out of specialized agriculture. This suggeststhat diversification into other activities might be a mechanism that house-holds use to cope with shocks that affect agricultural production.2 We do notfind any evidence that the wealth of the household is a determining factor.The key motivation appears to be income related (lower incomes) and incomeshocks (losses to income).

There is no evidence that the characteristics of the household head areimportant in determining the transition out of agriculture with the exceptionof ethnicity. We find that even when income differences are controlled for,ethnic minority households are more likely to transition out of agriculture.The proportion of ethnic minorities involved in specialized agriculture fellfrom around one-half in 2008 to only one-quarter in 2014. It should be notedthat a greater proportion of ethnic minorities remain in specialized agriculturein 2014 compared with Kinh households.3 A more detailed analysis of ethnicminority households is provided in Chapter 13.

Before exploring the pattern of diversification further, we consider brieflythe characteristics of the households that remain specialized in agriculture.Performing a similar analysis to that presented in Table 5.9, we find that olderhouseholds and ethnic minorities are significantly more likely to remainspecialized (results not shown). This implies, as suggested in Table 5.9, thatwhile ethnic minority households are more likely to transition out of agricul-ture they are still more likely than Kinh households to remain specialized. Wealso find that households that remain specialized are less likely to suffer fromnatural and economic shocks, providing further evidence that diversificationappears to be a push factor for vulnerable households. There is no evidence to

2 See Wainwright, Tarp, and Newman (2012) for a full analysis of the role of diversification inhelping households to manage risks using the VARHS data.

3 This analysis focuses on waged employment and enterprise operation as the two maindiversification activities undertaken by households. The results highlight that ethnic minorityhouseholds are more likely to remain specialised in agriculture; however diversification intoother areas is also possible. Approximately 80 per cent of ethnic minority households earnincome from common property resources compared to only 23 per cent of Kinh households.Looking at the proportion of income earned by ethnic minorities from different sources,however, shows that approximately 50 per cent of income comes from agriculture and 8 per centfrom common property resources, revealing that these households are still highly reliant onagriculture.


105

suggest that remaining specialized is associated with income, wealth, or otherhousehold characteristics.To explore this further, we consider whether there are certain household

characteristics associated with moving from agriculture into different types ofactivities. Conditioning on households that transition out of agriculture, weexplore the factors that determine diversification into labour (column (2)),household enterprises (column (3)), and labour with a household enterprise(column (4)). In each of these cases, some agricultural activities are main-tained by the households. In column (5) we consider the factors that deter-mine the full transition out of agriculture into other activities.

Table 5.9 Determinants of the transition out of agriculture

(1) (2) (3) (4) (5)

Transitionedout of ag.

Diversified intolabour & ag.

Diversifiedinto enterprise& ag.

Diversified intolabour, enterprise,& ag.

Diversified intoother activities(no ag.)

Lag log(income)

�0.029*** �0.025 0.004 �0.039*** 0.055***

(0.008) (0.017) (0.019) (0.015) (0.020)Lag asset index 0.001 0.007 �0.008 �0.001 0.004

(0.007) (0.016) (0.016) (0.013) (0.018)Lag HH size �0.021*** �0.035*** 0.002 �0.017 0.052***

(0.006) (0.013) (0.013) (0.012) (0.015)Lag femalea �0.005 0.068 0.002 �0.048 �0.071

(0.018) (0.053) (0.046) (0.037) (0.053)Lag marrieda �0.007 0.029 �0.009 �0.039 �0.060

(0.019) (0.051) (0.040) (0.035) (0.056)Lag agea 0.001 �0.002 0.000 �0.001* 0.002*

(0.001) (0.001) (0.001) (0.001) (0.001)Lag higher ed.a 0.012 �0.006 0.007 0.021 0.032

(0.014) (0.040) (0.026) (0.025) (0.039)Lag children �0.009 0.020 0.018 �0.005 �0.012

(0.018) (0.034) (0.028) (0.036) (0.046)Lag ethnicminority

0.151*** �0.022 0.038 �0.024 0.026

(0.019) (0.045) (0.026) (0.022) (0.036)Natural shock 0.028*** 0.001 �0.023 0.015 �0.001

(0.010) (0.025) (0.018) (0.018) (0.024)Economic shock 0.027** 0.004 0.031 �0.049*** 0.020

(0.011) (0.026) (0.024) (0.018) (0.030)Time dummies Yes Yes Yes Yes YesHouseholdspecificmeans

Yes Yes Yes Yes Yes

Number ofhouseholds

2,150 630 630 630 630

Number ofobservations

6,174 1,098 1,098 1,098 1,098

Notes: a Refers to characteristic of the head of household. Each specification is estimated using a random effectsestimator. Standard errors clustered at the household level are presented in parentheses. *** indicates significance atthe 1% level, ** indicates significance at the 5% level, * indicates significance at the 10% level.



106

Themain driving factors behind which activities households that transitionengage in, are income related. Lower income households are more likely totransition into waged employment, while income does not appear to be afactor in making the transition to a household enterprise. Higher incomehouseholds are more likely to make the full transition out of agriculture intoother activities. Overall, it is clear that the income level of households is themain determinant of the transition from specialized agriculture and the typesof activities into which households transition.

5.5 Conclusions

In this chapter we documented the extent to which structural transformationis observed at the microeconomic level through household-level income diver-sification. The VARHS data confirm the macroeconomic story. We observe asignificant shift in the allocation of labour from agriculture towards operatinga household enterprise and engaging in waged labour outside the home.

We find that diversified households have higher per capita consumptionmeasures than non-diversified households. In particular, households with anenterprise tend to have higher welfare (by about 20 per cent). We also exam-ined the welfare impact of the transition out of agriculture. Controlling forhousehold characteristics, initial income and wealth, we find that householdsthat moved from specialized agriculture between 2008 and 2014 experiencedwelfare gains of the order of 13 per cent. Those that transitioned into wagedlabour experienced gains of around 15 per cent, while those that transitionedinto both waged labour and a household enterprise experienced gains ofaround 23 per cent.

In the final part of our analysis we explore what factors drive the decision ofhouseholds to transition out of agriculture. We find that the decision isprimarily income related. Low-income households are more likely to makethe transition, as are households that have experienced income shocks. Also ofnote is the fact that ethnic minority households are much more likely totransition out of specialized agriculture. Only the richest households, how-ever, completely abandon agricultural production.

While agriculture remains the main source of income and employment forthe vast majority of rural Vietnamese, our results strongly confirm that diver-sification is happening on a large scale in Viet Nam. This process will continueand is likely to accelerate. Our core finding is that diversification has been, onaverage, welfare improving. While the most beneficial form of diversificationis into a household enterprise, there are many other factors that determine thesuccess or otherwise of entrepreneurial activities in rural settings (Kinghanand Newman 2015). These include access to finance, education, market


107

access, and others. Future research is needed to understand the relativeimportance of these factors in cultivating enterprises, especially given thefocus in Viet Nam’s reform agenda up to 2035 of developing a facilitatingenvironment for enterprise operation (World Bank 2016). Diversification intowaged employment is also an important source of welfare gain in our analysis,leading to welfare improvements of around 15 per cent. As such, close atten-tion should be paid to job creation, particularly in rural areas, for those leavingagricultural production.

Appendix

Table 5.A1 List of industry sectors of enterprise operation

Industry Freq. %

Agriculture, forestry, & aquaculture 179 7.8Mining & quarrying 9 0.39Processing & manufacturing 723 31.49Water & waste management 20 0.87Construction & engineering 33 1.44Wholesale & retail trade 638 27.79Transport & storage 93 4.05Accommodation & food services 207 9.02Information & communication 7 0.3Financial, banking, insurance, & real estate 5 0.22Professional, scientific, & technical 38 1.66Admin. & support services 21 0.91Education & training 2 0.09Health care 13 0.57Arts, entertainment, & recreation 94 4.09Other service activities 214 9.32Total 2,296 100

Source: Authors’ compilation based on survey data comprised from VARHS for the years 2008 to 2014.


108

Table 5.A2 List of industry sectors of external employment

Industry Freq. %

Agriculture, forestry, & aquaculture 915 17.23Mining & quarrying 52 0.98Processing & manufacturing 998 18.79Water & waste management 11 0.21Construction & engineering 1,274 23.98Wholesale & retail trade 127 2.39Transport & storage 162 3.05Accommodation & food services 123 2.32Information & communication 40 0.75Financial, banking, insurance, & real estate 35 0.66Professional, scientific, & technical 75 1.41Admin. & support services 84 1.58Education & training 290 5.46Political organizations 484 9.11Health care 132 2.48Arts, entertainment, & recreation 38 0.72Other service activities 472 8.89Total 5,312 100


Table 5.A3 Summary statistics

2008 2010 2012 2014

Mean Std dev. Mean Std dev. Mean Std dev. Mean Std dev.

Log food exp. p.c. 5.43 0.89 5.63 0.73 5.90 0.71 5.82 0.69Log income 10.75 0.87 10.91 0.90 11.05 0.85 11.13 0.87Asset index 0.04 1.08 0.19 1.09 0.32 1.08 0.36 1.08HH size 4.56 1.77 4.34 1.73 4.23 1.79 4.14 1.80Female 0.21 0.41 0.21 0.41 0.22 0.41 0.24 0.43Married 0.82 0.38 0.82 0.39 0.79 0.41 0.78 0.41Age 40.24 11.84 41.89 12.46 43.23 13.09 45.62 13.21Higher education 0.16 0.37 0.19 0.39 0.18 0.39 0.21 0.41Children 0.49 0.50 0.53 0.50 0.49 0.50 0.47 0.50Ethnic minority 0.21 0.40 0.20 0.40 0.20 0.40 0.20 0.40Natural shock 0.43 0.50 0.43 0.49 0.32 0.47 0.24 0.43Economic shock 0.23 0.42 0.17 0.37 0.19 0.39 0.14 0.34



109

Table 5.A4 Impact of diversification on household welfare, results for control variables

Table 5.7 Column (2) Table 5.7 Column (3)

Asset index 0.092*** 0.094***(0.014) (0.015)

HH size �0.131*** �0.131***(0.013) (0.013)

Female 0.001 0.002(0.079) (0.082)

Married 0.061 0.065(0.068) (0.070)

Age 0.002 0.003(0.002) (0.002)

Higher education 0.008 0.006(0.037) (0.038)

Children 0.051* 0.064**(0.031) (0.032)

Ethnic minority �0.137 �0.113(0.144) (0.157)

Natural shock �0.001 �0.005(0.019) (0.021)

Economic shock 0.013 0.012(0.023) (0.026)

L.log income �0.019(0.014)

L.Asset index 0.019(0.012)

L.HH size 0.013(0.012)

L.Female �0.006(0.073)

L.Married �0.050(0.061)

L.Age �0.001(0.002)

L.Higher education �0.004(0.032)

L.Children �0.072**(0.033)

L.Ethnic minority 0.041(0.112)

L.Natural shock �0.003(0.020)

L.Economic shock 0.001(0.024)

Notes: Each specification includes household fixed effects. Standard errors clustered at the household level are presentedin parentheses.*** indicates significance at the 1% level,** indicates significance at the 5% level,* indicates significance atthe 10% level.



110

Table 5.A5 Impact of diversification out of agriculture on household welfare, results forcontrol variables


Activities of non-transition hhs:Ag. & labour 0.076 0.078*

(0.047) (0.047)Ag. & ent. 0.132** 0.135**

(0.057) (0.057)Ag., labour, & ent. 0.224*** 0.227***

(0.054) (0.054)Labour only 0.020 0.022

(0.080) (0.080)Ent. only 0.187** 0.188**

(0.082) (0.082)Labour & ent. 0.097 0.100

(0.081) (0.081)

Household characteristics:Asset index 0.094*** 0.093***

(0.015) (0.015)HH size �0.130*** �0.131***

(0.013) (0.013)Female 0.003 0.003

(0.082) (0.082)Married 0.065 0.066

(0.069) (0.070)Age 0.003 0.003

(0.002) (0.002)Higher education 0.007 0.007

(0.038) (0.038)Children 0.064** 0.066**

(0.032) (0.032)Ethnic minority �0.122 �0.122

(0.158) (0.158)Natural shock �0.003 �0.004

(0.021) (0.021)Economic shock 0.012 0.013

(0.026) (0.026)L.log income �0.020 �0.019

(0.014) (0.014)L.Asset index 0.019 0.019

(0.012) (0.012)L.HH size 0.013 0.014

(0.012) (0.012)L.Female �0.005 �0.001

(0.073) (0.073)L.Married �0.048 �0.047

(0.060) (0.060)L.Age �0.0001 �0.0001

(0.002) (0.002)L.Higher education �0.004 �0.005

(0.032) (0.032)L.Children �0.073** �0.073**

(0.033) (0.033)L.Ethnic minority 0.044 0.044

(0.11) (0.11)

(continued )


111

References

Ackah, C. (2013). ‘Nonfarm Employment and Incomes in Rural Ghana’. Journal ofInternational Development, 25(3): 325–39.

Benedikter, S., G. Waibel, S. Birtel, and B. T. Tran (2013). Local Entrepreneurship in VietNam’s Rural Transformation: A Case Study from the Mekong Delta. Bonn: Centre forDevelopment Research (ZEF), Can Tho City Institute for Socio-Economic Develop-ment Studies (CIDS), Southern Institute of Social Sciences (SISS).

Bezu, S. and C. Barrett (2012). ‘Employment Dynamics in the Rural Nonfarm Sector inEthiopia: Do the Poor Have Time on their Side?’ Journal of Development Studies, 48(9):1223–40.

Bezu, S., C. B. Barrett, and S. T. Holden (2012). ‘Does the Nonfarm Economy OfferPathways for UpwardMobility? Evidence from a Panel Data Study in Ethiopia’.WorldDevelopment, 40(8): 1634–46.

Birthal, P. S., D. S. Negi, A. K. Jha, and D. Singh (2014). ‘Income Sources of FarmHouseholds in India: Determinants, Distributional Consequences and Policy Impli-cations’. Agricultural Economics Research Review, 27(1): 37–48.

Development Analysis Network (2003). ‘Off-Farm and Non-Farm Employment inSoutheast Asia Transitional Economies and Thailand’. Available at: <http://www.cdri.org.kh/webdata/download/dan/ddan3.pdf>. Accessed 9 August 2016.

Economica Vietnam (2013). ‘The Non-Farm Household Business Sector in VietNam’. February. Available at: <http://www.economica.vn/FormDoc/tabid/192/topic/H68T30279650371456/ language/en-US/Default.aspx>. Accessed 9 August 2016.

Giesbert, L. and K. Schindler (2012). ‘Assets, Shocks and Poverty Traps in Rural Mozam-bique’. World Development, 40: 1594–609.

Haggblade, S., P. Hazell, and T. Reardon (2010). ‘The Rural Non-Farm Economy: Pros-pects for Growth and Poverty Reduction’. World Development, 38(10): 1429–41.

Hoang, T. X., C. S. Pham, and M. A. Ulubasoglu (2014). ‘Non-Farm Activity, HouseholdExpenditure, and Poverty Reduction in Rural Viet Nam: 2002–2008’. World Develop-ment, 64: 554–68.

Imai, K. S., R. Gaiha, and G. Thapa (2015). ‘Does Non-Farm Sector Employment ReduceRural Poverty and Vulnerability? Evidence from Viet Nam and India’. Journal of AsianEconomics, 36(2): 47–61.

Table 5.A5 Continued


L.Natural shock �0.003 �0.002(0.020) (0.020)

L.Economic shock 0.003 0.001(0.0247) (0.024)

Notes: Each specification includes household fixed effects. Standard errors clustered at the household level are presentedin parentheses. *** indicates significance at the 1% level, ** indicates significance at the 5% level, * indicates significanceat the 10% level.



112

http://www.cdri.org.kh/webdata/download/dan/ddan3.pdf

http://www.cdri.org.kh/webdata/download/dan/ddan3.pdf

http://www.economica.vn/FormDoc/tabid/192/topic/H68T30279650371456/language/en-US/Default.aspx

http://www.economica.vn/FormDoc/tabid/192/topic/H68T30279650371456/language/en-US/Default.aspx

Kinghan, C. and C. Newman (2015). Social Capital, Political Connections and HouseholdEnterprises: Evidence from Viet Nam. WIDER Working Paper 2015/001. Helsinki:UNU-WIDER.

Lanjouw, P., R. Murgai, and N. Stern (2013). ‘Non-Farm Diversification, Poverty, Eco-nomic Mobility, and Income Inequality: A Case Study in Village India’. AgriculturalEconomics, 44(4–5): 461–73.

Meyer, B. D. and J. X. Sullivan (2011). ‘Viewpoint: Further Results on Measuringthe Well-Being of the Poor Using Income and Consumption’. Canadian Journal ofEconomics/Revue Canadienne d’Economique, 44(1): 52–87.

Micevska, M. (2008). ‘Rural Non-Farm Employment and Incomes in the Himalayas’.Economic Development and Cultural Change, 57(1): 163–93.

Olugbire, O. O., A. O. Falusi, A. I. Adeoti, and A. S. Oyekale (2012). ‘Determinants ofNon-Farm Income Diversification among Rural Households in Nigeria’. Journal ofAmerican Science, 8(1): 77–82.

Oostendorp, R. H., T. Q. Trung, and N. T. Tung (2009). ‘The Changing Role of Non-Farm Household Enterprises in Viet Nam’. World Development, 37(3): 632–44.

Porter, C. (2012). ‘Shocks, Consumption and Income Diversification in Rural Ethiopia’.Journal of Development Studies, 48(9): 1209–22.

Vijverberg,W. P. and Haughton, J. H. (2002).Household Enterprises in Viet Nam: Survival,Growth, and Living Standards. World Bank Publications Working Paper 2773. WorldBank Publications.

Wainwright, F., F. Tarp, and C. Newman (2012). ‘Risk and Household Investment Deci-sions: Evidence from Rural Viet Nam’. Paper presented at the European EconomicsAssociation Conference, Malaga, Spain.

World Bank (2016). Viet Nam 2035: Toward Prosperity, Creativity, Equity and Accountabil-ity. Washington DC: World Bank.


113

Part IIKey Production Factors and Institutions

6

Land Issues

Markets, Property Rights, and Investment

Thomas Markussen

6.1 Introduction

The transfer of agricultural land use rights from collectives to individualhouseholds in 1988was a key element of the DoiMoi reforms. In 1993, privateland property rights were further strengthened as a massive programme ofsystematic land titling was initiated and land holders gained the rights to sell,rent, exchange, mortgage, and bequest their plots. These developments areoften credited as an important driver of rural economic growth in Viet Nam(for example, Pingali and Xuan 1992; Rozelle and Swinnen 2004; Deiningerand Jin 2008; Do and Iyer 2008; Newman, Tarp, and van den Broeck 2015). Onthe other hand, the literature also documents that household land propertyrights are far from complete and not always well protected. For example,Markussen, Tarp, and van den Broeck (2011) point out that many householdsface binding restrictions on crop choice, Anderson and Davidsen (2011) showthat land titling is perceived to be severely affected by corruption, andMarkussen and Tarp (2014) show that the risk of government land expropri-ation is significant and depends on whether or not a household has informalties with local government officials. Khai and colleagues (2013) documentthat, while land market transactions seem to increase efficiency as well asequity of land use, land markets are still very thin in many areas of Viet Nam.

This chapter investigates land issues in Viet Nam from different angles. TheViet Nam Access to Resources Household Survey (VARHS) includes a highlydetailed land module. In contrast with most other large-scale economic sur-veys, it collects panel data not only at the household but also at the land plotlevel. This is highly useful in some of the analyses I conduct.

The chapter first considers access to agricultural land by reporting theshare of landless rural households (Section 6.2).1 It then turns to analysingissues of farm size and land fragmentation (Section 6.3), land sales and rentalmarkets (Section 6.4), and property rights to land (Section 6.5). While thefirst five sections are mainly descriptive, Section 6.6 presents fixed effectsregressions at the plot level to determine the causal effect of land titles onagricultural investment. This section demonstrates a significant and strongeffect of land titling on household investment in irrigation. Remarkably,this effect is only present in upland regions, where titling is least prevalent.The policy implication is that titling should be expanded in the highlands.Section 6.7 concludes.

6.2 Landlessness

I first consider landlessness, a phenomenon often associated with poverty andvulnerability in developing countries. Households that neither own nor oper-ate any agricultural land are defined as ‘landless’. Figure 6.1 (panel a) showsthe prevalence of landlessness over time and in five different regions. Thefigure shows that landlessness in the 2006–14 VARHS panel is low (around8 per cent in 2014) and relatively stable over time.2 There is significantvariation across regions. Landlessness is highest in the Mekong River Delta(MRD) and, in 2014, the Central Coast (12–18 per cent), and lowest in the RedRiver Delta and the North (around 3–6 per cent). There is a tendency towardsconvergence over time, driven by a moderate increase in landlessness in thenorthern parts of Viet Nam and a moderate drop in the Southern and CentralLowlands. The increase in landlessness in the Red River Delta might be drivenby improving off-farm opportunities in and around Hanoi. It is unclear whatexplains the large drop in landlessness in the MRD between 2012 and 2014.Figure 6.1 (panel b) shows landlessness by income quintile. Since a compre-

hensive measure of income could not be computed for 2006, results for thatyear are not included. The figure shows clear and stable differences acrossincome groups, but not in the direction one might expect. Landlessness ishighest in the richest quintile (around 12 per cent) and lowest in the poorestquintile (around 5 per cent). Hence, landlessness is not generally associatedwith poverty in Viet Nam. This is, of course, partly explained by the patternsrevealed in Figure 6.1; landlessness is most prevalent in the Southern and

1 Other overviews of land issues in Viet Nam include Kerkvliet (2006); Brandt (2006); andMarsh,MacAulay, and Hung (2006).

2 Using the full, representative VARHS sample, landlessness in 2014 is 11 per cent (the same as in2008). Hence, with the full sample, the slight downward trend in the panel sample between 2008and 2014 is not present.

Key Production Factors and Institutions

118

Central Lowlands, which are also relatively rich regions. However, the positiveassociation between landlessness and income is present within each region,except the Central Highlands, where there is no clear correlation betweenincome and landlessness (results not shown). Therefore, results are consistentwith the findings in Ravallion and van de Walle (2008), who argue thathouseholds in Viet Nam typically do not become landless because they areexposed to negative, economic shocks, but rather sell their land in order totake up new opportunities in the growing non-farm economy.The background for low levels of landlessness and the absent correlation

between landlessness and poverty is, of course, the highly egalitarian landreforms initiated in 1988, which in turn were premised on the collectivization

Panel a: By region

0

2

4

6

8

10

12

14

16

18

20

2006 2008 2010 2012 2014

Perc

enta

ge

of

lan

dle

ssPe

rcen

tag

e o

f la

nd

less

Year

2

4

6

8

10

12

14

16Panel b: By income quintile

2008 2010 2012 2014Year

Red River Delta North Central Coast Central Highlands Mekong River Delta All

Richest quintile Second-richest Middle Second-poorest Poorest quintile

Figure 6.1 LandlessnessNotes: N = 2,162 (Observed in each year, i.e. there are 5 x 2,162 = 10,810 observations). Onlyhouseholds that neither own nor operate any agricultural land are defined as landless.Source: Author’s calculations based on VARHS 2006–14.

Land Issues

119

of agriculture after the communist revolution (Ravallion and van de Walle2004, 2006). Equality of the agricultural land distribution is arguably one ofthe most important, positive aspects of the communist legacy in Viet Nam.

6.3 Farm Size and Land Fragmentation

An implication of egalitarianism in land distribution, combined with high,rural population density, is that Vietnamese farms are small. In addition,each farm tends to be divided into lots of separate plots, especially in theNorth (see Markussen et al. 2013). This section considers developmentsin farm size and land fragmentation between 2006 and 2014. Figure 6.2(panel a) shows median farm size (defined as operated agricultural area) byregion. The figure documents that farms are much bigger in the CentralHighlands than in other regions and significantly larger in southern than innorthern areas. The latter difference (between North and South) has longhistorical roots. A part of the background is, of course, the longer period ofcommunist rule in the North, which meant that agricultural collectivizationwas much more comprehensive in the North than in the South. This in turnled to a more egalitarian, post-Doi Moi land distribution in the North. In theSouth, many households simply continued to farm the land they had farmedbefore the communist takeover. The North–South differences go back evenlonger than communism, though. Since pre-colonial times, population dens-ity was significantly higher in the Red River Delta than in the MRD, meaningthat landholdings per household were significantly lower in Red River Delta(Gourou 1936; Popkin 1979).The figure shows a moderate decrease in median farm size over time (from

around 3,700 m2 to around 3,250 m2 significant at the 1 per cent level in amedian regression).3 While farms in the VARHS panel are getting smaller inmost regions, they are actually growing in the Central Highlands, implying atendency towards interregional divergence, since farms in the Central High-lands were already larger than in other regions in 2006. A possible explanationis return to scale in commercialized agriculture, in particular coffee produc-tion, which is prevalent in the Central Highlands.Figure 6.2 (panel b) shows median farm size by income quintile. The results

show that farms are biggest in the poorest quintile. Among the four richestquintiles, there is no strong association between income and farm size. Thisagain shows that there is no straightforward association between povertyand access to land in Viet Nam. It is, of course, important to remember that

3 The decrease in median farm size is somewhat stronger when the full, representative VARHSsample is used (from 3,700 m2 in 2006 to 3,050 m2 in 2014).


120

the figure does not account for the quality of land and that land in thehighlands is often of lower quality than in the lowlands. The incidence ofpoverty is also significantly higher in the mountains than in the plains.Another key factor behind the results, however, is the importance of therural non-farm economy. Most households have other sources of incomethan agriculture, and non-farm employment is often more remunerativethan farming (cf. Chapter 5).

While Figure 6.2 considers ‘inter-farm land fragmentation’ (the division ofland between many, relatively small farms), Figure 6.3 presents results on

0

2000

4000

6000

8000

10000

12000

14000

16000

Panel a: By region

2006 2008Year

Year

2010 2012 2014

2000

2500

3000

3500

4000

4500

5000

Panel B: By income quintile

2008 2010 2012 2014

Med

ian

op

erat

ed f

arm

area

(sq

m)

Med

ian

op

erat

ed f

arm

area

(sq

m)

Red River Delta North Central Coast Central HighlandsMekong River Delta All


Figure 6.2 Farm sizeNotes: N = 1,953 households in 2006 (slightly less in later years). Farm size is defined as operatedrather than owned area (i.e. plots rented in are included and plots rented out are excluded). Onlyhouseholds operating agricultural land are included.Source: Author’s calculations based on VARHS 2006–14.

Land Issues

121

‘intra-farm fragmentation’ (the division of each farm into separate plots).The figure shows the average number of agricultural plots operated by farm-ing households. Intra-farm land fragmentation is potentially problematicbecause it prevents the use of large-scale machinery and uses land resourcesfor boundary demarcation and labour resources for travelling between plots.The Viet Nam government has aimed to reduce land fragmentation byimplementing land consolidation programmes in many communes, espe-cially northern areas. These programmes aim to consolidate land holdings byfacilitating plot exchanges between households. Figure 6.3 suggests thatthese programmes may have had some effect. The mean number of plotsoperated has dropped from 5.8 in 2006 to 4.1 in 2014, with the sharpestdecrease recorded in the Red River Delta (from 6.6 to 4.0). An alternativeinterpretation is that, because the panel households are getting older, theyoperate fewer plots (i.e. rent fewer plots in, rent more plots out, and passmore plots on to younger relatives). However, there is also a significantdecline in the number of plots owned (rather than operated), from 5.7 in2006 to 4.1 in 2014. The rate of giving plots away (for example, as bequests)is stable over time (4.4 per cent of households gave at least one plot awayduring the two years before the 2006 survey; the equivalent number for the2014 survey is 4.3 per cent). Hence, the results do suggest that intra-farmland consolidation is taking place.In sum, while there is little evidence of inter-farm land consolidation (if

anything, farms are getting a little smaller), there is evidence of moderateprogress towards intra-farm consolidation.

0

Mea

n n

umb

er o

f p

lots

op

erat

edp

er f

arm

ing

HH

1

2

3

4

5

6

7

8

2006 2008 2010Year

2012 2014


Figure 6.3 Number of plots operated, by regionNotes: N = 1,953 households in 2006 (slightly less in later years). Only households operatingagricultural land are included.Source: Author’s calculations based on VARHS 2006–14.


122

6.4 Land Markets

In a dynamic economy such as Viet Nam’s, where new economic opportun-ities constantly arise, it is of high importance that land can be shifted betweendifferent users without excessive friction. Therefore, well-functioning landmarkets are essential. This section considers participation rates in land salesand rental markets.

In contrast with China, agricultural land sales markets are legal in Viet Nam.Legality is not a sufficient condition for activity, however. As documented inKhai and colleagues (2013), land market sales have played a relatively minorrole for land allocation in large parts of Viet Nam, particularly in the North.On aggregate, only about 8 per cent of plots operated by households havebeen acquired through purchase (63 per cent have been given by the state,15 per cent were received as inheritance, and 13 per cent were cleared byhouseholds). In the North, only about 2.5 per cent of plots were acquiredthrough the market (compared with 11 per cent in the Southern Lowlands and46 per cent in the Central Highlands). Part of the reason for low levels of activityin the landmarket is the relativelyhighdegreeof efficiency that characterized theadministrative land allocation implemented after 1988 (Ravallion and van deWalle 2004). In addition,however, land saleshave, until recently, been subject toa virtual taboo in large parts of Northern Viet Nam, where land sales marketsnever existed in the past, even before the rise of communism (Popkin 1979).

Figure 6.4 shows the share of households that purchased (panel a) or sold(panel b) agricultural land during the two years prior to each survey round, byregion. Results document clearly that land sales markets are much more activein theCentralHighlands than in any other region. TheCentralHighlands differfrom other regions in the sense that recent decades have seen massive inwardmigration and changes in agricultural activities. The large increase in coffeeproduction is themost important of these changes. Therefore, land allocation ismuchmore dynamic in the Central Highlands than elsewhere, asmigrants andother residents attempt to adapt land holdings to changing circumstances.Figure 6.4 also show that participation rates in land sales markets tend to behigher in the MRD than in northern and central areas, although activity levelsare much lower in the MRD than in the Central Highlands.

Overall, activity levels are largely stable over time. Sharp increases in activitylevels are recoded in the Central Highlands in 2008 (purchases) and 2014(sales). The reasons behind these specific developments are unclear.

Figure 6.5 presents activity levels in land sales markets by income quintile.Results show clearly that the richest households are most active on thesupply as well as the demand side of the market. Hence, there is no evidencethat land sales markets increase inequality, in the sense of transferringland from the poor to the rich. However, it is a concern that markets mainly

Land Issues

123

serve the better-off part of the population, leaving many poorer householdsexcluded. In terms of land sales, there is a tendency towards convergencebetween income groups over time, but no such trend is apparent in the caseof land purchases.I now turn to considering the land rental market. Figure 6.6 shows, respect-

ively, the share of households renting land in and out, by region. Results showthat the regional pattern is quite different for rental than for salesmarkets. The

Mea

n n

umn

er o

f p

lots

op

erat

edp

er f

arm

ing

HH

Year

0

2

4

6

8

10

12

2006 2008 2010 2012 2014

Red River Delta North Central Coast Central Highlands

Mekong River Delta All

Panel b: Sales

Panel a: Purchases

0

2

4

6

8

10

12

14

16

2006 2008 2010

Year

2012 2014Perc

enta

ge

of

HH

wh

o b

oug

ht

lan

din

th

e la

st 2

yea

rs


All MvRiver Delta

Figure 6.4 Land purchases and sales in the last two years, by regionNotes: Panel a: N = 2,025 households in 2006 (slightly less in later years). Only land-owninghouseholds included. Panel b: N = 2,162 households.Source: Author’s calculations based on VARHS 2006–14.


124

most active region is the Red River Delta, followed by the Central Coast. TheNorth is the least active region in the case of renting in, while the CentralHighlands are least active in terms of renting out.4 Hence, in the Red RiverDelta, low activity of land sales markets are, to a large extent, compensated forby high activity in rental markets. In the North, however, this is not the case.

There are clear time trends: the share of households renting land in isdecreasing, while the share renting land out is increasing. These opposite

0Perc

enta

ge

of

HH

wh

o b

oug

ht

lan

din

th

e la

st 2

yea

rsPe

rcen

tag

e o

f H

H w

ho

so

ld la

nd

in t

he

last

2 y

ears

1

2

3

4

5

6

Panel a: Purchases

Panel b: Sales

2008 2010Year

Year

2012 2014

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

2008 2010 2012 2014



Figure 6.5 Land purchases and sales in the last two years, by income quintileNotes: Panel a: N = 1,938 households in 2008 (slightly more in later years). Only land-owninghouseholds included. Panel b: N = 2,162 households.Source: Author’s calculations based on VARHS 2008–14.

4 Rates of renting in and out may differ for several reasons. First, the same landlord may rent outto several tenants, and vice versa. Second, landlords may not be households, but rather communeauthorities or corporate entities, which are not captured by the survey.

Land Issues

125

trends are likely to result from the ageing of panel households. Rental activ-ities are strongly correlated with age of the household head (younger house-holds rent land in, older households rent out). However, the upward trend inrenting out (10 percentage points) is markedly stronger than the downwardtrend in renting in (3 percentage points), suggesting that overall activity levels

0Perc

enta

ge

of

HH

ren

tin

g la

nd

out

Perc

enta

ge

of

HH

ren

tin

g la

nd

out

5

10

15

20

25

30

35

40

Panel a: Renting in

Panel b: Renting out

2006 2008 2010Year

Year

2012 2014

0

5

10

15

20

25

30

2006 2008 2010 2012 2014





Figure 6.6 Share of households renting land in and out, by regionNotes: Panel a: N = 1,953 households in 2006 (slightly less in later years). Only householdsoperating agricultural land included. Panel b: N = 2,015 households in 2006 (slightly less in lateryears). Only land-owning households included.Source: Author’s calculations based on VARHS 2006–14.


126

in land rentalmarkets have increased. Indeed, the share of households involvedin land rentals on at least one side of the market increased from 28 per cent in2006 to 34 per cent in 2014 (a highly statistically significant change).

Figure 6.7 shows rental market participation by income quintile. Results arequite interesting. There is no clear correlation between income and participa-tion rates on the demand side (panel a). The poorest and the richest quintiles

10

12

14

16

18

20

22

24

26

Panel a: Renting in

Panel b: Renting out

2008 2010Year

Perc

enta

ge

of

HH

ren

tin

g la

nd

Perc

enta

ge

of

HH

ren

tin

g la

nd

2012 2014

0

5

10

15

20

25

30

35

2008 2010 2012 2014

Richest quintile Second-richest Middle

Middle

Second-poorest Poorest quintile

Year

Richest quintile Second-richestSecond-poorest Poorest quintile

Figure 6.7 Share of households renting land in and out, by income quintileNotes: Panel a: N = 1,876 households in 2008 (slightly deviations from this in later years). Onlyhouseholds operating agricultural land included. Panel b: N = 1,938 households in 2008 (slightlymore in later years). Only land-owning households included.Source: Author’s calculations based on VARHS 2008–14.

Land Issues

127

are the two least active groups, with the three middle quintiles all beingsomewhat more active. On the supply side (panel b), however, there is avery clear tendency towards higher participation among richer households.This suggests that land rental markets are ‘progressive’ in the sense of trans-ferring land from rich to poor households.5 These findings are consistent withthe results reported in Deininger and Jin (2008) and in Khai and colleagues(2013). Differences across income groups are stable over time.

6.5 Property Rights

The 1993 Land Law endowed landholders with a rather comprehensive set ofland rights. Land continues formally to be owned by the state (‘the People’,states the law), but land users gained twenty years’ use rights for plots desig-nated for annual crops land and fifty years for perennial crops land. Theserights were guaranteed through the issuance of land-use certificates (LUCs),which also imply the rights to sell, rent, mortgage, exchange, and bequest aplot of land. Land rights have been gradually strengthened and clarifiedthrough various revisions of Land Law. The 2013 Land Law (in effect from2014) extends the duration of use rights to fifty years for all types of land.While formal protection of tenure security and transfer rights is fairly strong,Markussen and Tarp (2014) show that de facto property rights are not com-plete. The actual risk of losing land against the will of the household issignificant in many areas. This section focuses on formal property rights toland; that is, LUCs.Figure 6.8 shows the share of land plots held with an LUC. Only plots

owned by households are included, that is, rented plots are excluded. Purelyresidential plots are also excluded. Results show that land titling is compre-hensive (covering around 80 per cent of plots) but not complete, and thatthe share of plots titled is approximately stable over the period studied. Thepattern of overall stability is the result of different, opposing forces. On theone hand, titling efforts are ongoing, although much less vigorously than inthe 1990s (cf. Do and Iyer 2008). On the other hand, plots may cease to betitled if they change hands through sale or inheritance and title documentsare not updated. An obvious barrier to registration of land transactions (andthereby titling) is the presence of informal fees in the land administrationsystem. Anderson and Davidsen (2011) show that corruption is perceived tobe widespread in the public land administration. Also, plots obtained by

5 In principle, an alternative explanation could be that richer households rent out to companies,rather than renting out to poor households. However, data on the identity of tenants show that lessthan 2 per cent of tenants on plots rented out were firms.


128

clearing the forest are often not titled. Only 42 per cent of plots obtainedthrough forest clearing are titled. Among plots cleared in the last five years,only 13 per cent are held with an LUC. This probably explains the downwardtrend in titling in the Northern Uplands, where forest clearing is most preva-lent. There is important, interregional variation in land titling. Titling is mostprevalent in the MRD, where almost all plots are held with an LUC. Titling is

30

40

50

60

70

80

90

100

Panel a: By region

Panel b: By income quintile

2006 2008 2010Year

Year

2012 2014

50

55

60

65

70

75

80

85

90

2006 2008 2010 2012

Perc

enta

ge

of

plo

ts w

ith

LU

CPe

rcen

tag

e o

f p

lots

wit

h L

UC

Northern lowlands Northern uplands Southern lowlandsCentral highlands All

Richest quintile Second-richest MiddleSecond-poorest Poorest quintile

Figure 6.8 Land-use certificatesNotes: Panel a: N = 9,910 plots in 2006 (slightly less in later years). Only plots owned by householdsincluded (i.e. plots rented in are excluded). Panel b: N = 9,422 plots in 2008 (somewhat less in lateryears). Only plots owned by households included (i.e. plots rented in are excluded).Source: Author’s calculations based on VARHS 2006–14.

Land Issues

129

least prevalent in the Central Highlands. There is significant heterogeneitywithin the North. Titling is very widespread in Phu Tho, a relatively rich,mostly lowland province. In the remote, highland provinces of Dien Bienand Lai Chau, on the other hand, only 41 and 46 per cent of plots, respect-ively, are titled.Panel b of Figure 6.8 shows land titling by income quintile of the plot

owner. There is a clear and stable income gradient in land titling, the preva-lence of LUCs being significantly lower in the poorest quintiles than in therichest. This is, of course, partly explained by the interregional patterndescribed in panel a of Figure 6.8 LUCs are least common in the NorthernUplands, which is also the poorest region. Whether weak property rights is acausal factor behind low income is not clear from these analyses (although theresults presented in Section 6.6suggest that it might be), but it is, in any case, acause for concern that the poorest segments of the population have theweakest, formal protection of property rights.

6.6 Effects of Land Titles on Agricultural Investment

Section 6.5 documented the fact that while a large share of plots are held withan LUC (‘titled’), titling is not complete, and varies significantly across regions.This section analyses the effects of variation in land rights. In particular,I investigate whether stronger property rights increase agricultural investment,focusing on two of the most important types of investment in Vietnamesefarming, namely irrigation and perennial crops.6

A large literature investigates the effects of land property rights on agricul-tural investment.7 These papers generally struggle to deal with an importantidentification problem, namely the potential effect of unobserved plot char-acteristics, which affect property rights (e.g. land titling) as well as investment.For example, households may own plots in the plains as well as in the hills(both types of landscape are, in many cases, present in the same community).Landmeasurement, border demarcation, and dispute resolution may be easierin the plains than in the hills, and a systematic titling programme, such as theprogramme implemented in Viet Nam from 1994 onward, might tend to focus

6 Most LUCs have been distributed through ‘systematic’ (supply-driven) titling programmes.Households with plots not covered by such programmes can apply for ‘sporadic’ (demand-driven)titling. However, this process involves considerable cost to the households, in part because ofcorruption in land administration (cf. Anderson and Davidsen 2011). This potentially explainswhy some households do not acquire titles even though they would benefit considerably fromhaving them.

7 See, e.g., Feder and Onchan (1987); Besley (1995); Braselle et al. (2002); Carter and Olinto(2003); Jacoby and Mansuri (2008); Do and Iyer (2008); Markussen (2008); Hornbeck (2010).


130

foremost on titling plots in the plains. At the same time, investment maydiffer systematically between the plains and the hills. For example, it might bemore feasible to invest in irrigation in the plains. Of course, one can try tocontrol for the factors that drive property rights and investment, but thisendeavour is likely to be only partly successful, as many potentially importantplot characteristics (soil type, quality of irrigation, etc.) are difficult to measureprecisely. In addition to using control variables, one might attempt to solveidentification problems through instrumental variables methods, as, forexample, in Besley (1995). As discussed in Markussen (2008), the validity ofthe instruments used for land property rights (for example, mode of plotacquisition) is, in most cases, uncertain.

The VARHS dataset offers unusual opportunities to deal with these issuesbecause the survey collects panel data not only at the household, but also atthe plot level. This allows us to track individual plots and see, for example,whether changes in property rights are associated with changes in invest-ment. Newman, Tarp, and van den Broeck (2015) exploit the plot panel toinvestigate the effects of LUCs on rice yields. They are particularly inter-ested in analysing effects of having both the husband’s and the wife’snames written into the LUC. They find that LUCs do indeed increaseproductivity and that this effect is not diminished by having two ratherthan one name in the LUC.8 The present analysis investigates one of thechannels through which property rights may affect rice yields, namelyinvestment in irrigation.

Given the prevalence of rice production in Vietnamese agriculture, theimportance of irrigation infrastructure is beyond dispute. In 2014, 73 percent of agricultural plots recorded in VARHS were irrigated (up from 68 percent in 2006). Investment in irrigation is conducted by the government aswell as by individual farmers and includes, for example, investment inreservoirs, canals, wells, dykes, and other water conservation infrastructure.I also consider investment in perennial crops. Since a number of yearstypically pass between planting and the first harvest of, say, coffee or man-goes, there is an important element of investment in the choice of perennialrather than annual crops. Households with high tenure security and access tocredit are more likely to plant perennial crops than others. Formalized prop-erty rights may improve access to credit as well as tenure security because aland title facilitates the process of using land as collateral for loans. In 2014,18 per cent of plots in VARHS were planted with perennial crops, up from15 per cent in 2006.

8 If husbands and wives have different objectives, shared property rights might have loweredinvestment and productivity relative to having only one person as the property rights holder.There was no evidence that this was the case.

Land Issues

131

I estimate plot level regressions models of the following type:

Ikpt ¼ β1LUCpt þ β2RESTRICpt þ β3Lpt þ θp þ γt þ εpt

where Ikpt is an indicator for the presence of investment good k on plot p inyear t. LUC is an indicator for the plot being held with an LUC. This is themain variable of interest. RESTRIC is an indicator for crop choice restrictions.On many plots, the choice of which crops to grow is restricted by land useplans. Most commonly, households are compelled to grow rice (Markussen,Tarp, and van den Broeck 2011). This indicator for crop choice restrictionsis mainly included as a control variable. Restrictions may affect investment,for example, because authorities are more likely to invest in irrigation forrestricted plots than for other plots. Also, restrictions and titling could becorrelated, for example, if systematic titling efforts are directed towardsrestricted plots. L is a measure of household labour resources (the number ofworking-age household members, with working age defined as 15–65 years).A higher labour force makes it more feasible to conduct investment projectsand may also increase the likelihood that households seek land titles, giventhat the title application process requires a certain amount of time and skills.The symbol θp is a plot fixed effect, equivalent to including a dummy variablefor each plot in the dataset; γt is a year fixed effect, which captures generaltrends in investment, as, for example, those arising from variation in nationaland international crop prices; and εpt is an error term. I allow errors to becorrelated within communes (the primary sampling unit of VARHS) but notacross. Only owned and operated agricultural plots are included. Some plotsare recorded with different areas in different years. This may reflect recordingerrors, or it may reflect real changes, as when a plot is expanded by clearing theforest, or by merging it with another plot. I exclude all plots with recordedchanges in area in order to avoid endogeneity problems. For example, if atitled plot is merged with a non-titled plot, the household may report that theinitially titled plot is not titled anymore. It may also change its report aboutthe investment status of the plot (e.g. if one of the merged plots has perennialcrops while the other does not).Table 6.1 presents results of estimating this model. All models contain plot

as well as year fixed effects. The first two regressions are models for a plotbeing, respectively, irrigated and planted with perennial crops. Results show astrong and statistically significant effect of LUCs on irrigation. Plots are morethan 6 percentage points more likely to be irrigated after they are titled thanbefore. In contrast, there is no effect of LUCs on perennial crops, contrary tothe findings in Do and Iyer (2008). One somewhat speculative explanation forthis negative finding is that perennial crops may function as a substitute forland titles. Trees and bushes are a visible and costly type of investment and


132

may in themselves strengthen a household’s claim to a plot of land, thusreducing the demand for an LUC (Besley 1995; Braselle, Gaspart, and Platteau2002). Crop choice restrictions have a positive effect on irrigation and anegative effect on perennial crops. This is not surprising since restrictionstypically compel the household to grow rice. Labour resources have a positiveeffect on investment in irrigation but no effect on perennial crops. The reasonfor the latter result might be that perennial crops often require less labour(after planting) than annual crops. Hence, the incentive to plant perennialcrops might be highest in households with scarce labour resources.

Regressions 3–6 further investigate the effect of LUCs on irrigation. A majorconcern is the potential importance of government investment in irrigationand the possibility that public irrigation investment might be correlated withtitling. The government mainly invests in canal infrastructure that bringswater to plots. Households, on the other hand, mainly conduct on-plotinvestment in wells, dykes, flattening, and so on. The dataset contains anindicator for a plot having ‘soil and water conservation infrastructure’. Sincethis reflects investment on the plot, it is likely to be driven mainly by house-hold activities. Regression 3 shows that plots are 5 percentage points morelikely to have soil and water conservation infrastructure after titling than

Table 6.1 Property rights and agricultural investment, plot-level regressions

Dependent variable:

Plotirrigated

Plot plantedwith perennialcrops

Plot has soil andwater conservationinfrastructure

Plotirrigatedfrom canals

Plotirrigatedfrom well

Plot irrigatedfrom spring,stream, or lake

(1) (2) (3) (4) (5) (6)LUC 0.064*** 0.0003 0.049** 0.030* 0.006 0.028*

(0.018) (0.006) (0.019) (0.016) (0.006) (0.015)Crop choice

Restricted0.124*** �0.022*** 0.124*** 0.139*** �0.003 �0.012

(0.012) (0.005) (0.013) (0.016) (0.004) (0.011)Working-age

HHmembers,log

0.040** �0.007 0.012 0.029 �0.003 0.014

(0.017) (0.010) (0.018) (0.019) (0.008) (0.016)

Plot fixedeffects

Yes Yes Yes Yes Yes Yes

Year fixedeffects

Yes Yes Yes Yes Yes Yes

Observations 30,125 29,409 30,001 30,125 30,125 30,125

Notes: Level of analysis: Plot. Linear probability models. Standard errors in brackets. Standard errors adjusted forcommune level clustering. Only plots with constant area included. *** p<0.01, ** p<0.05, * p<0.1.

Source: Author’s calculations based on VARHS 2006–14.

Land Issues

133

before (a statistically significant effect). This suggests that the effect of titling isnot driven by government investment activities. Regressions 4 to 6 providefurther support for this view. These regressions model the presence of irriga-tion from three different types of sources: (a) canals; (b) wells; and (c) springs,streams, or lakes. Only the first type of irrigation is likely to be substantiallydriven by government investment. Indeed, LUCs do have a significant, posi-tive effect on irrigation from canals. This may partly reflect governmentinvestment. However, LUCs also have a positive, significant effect on irriga-tion from springs, streams, or lakes. This is much more likely to be driven byhousehold investment in water conservation infrastructure.9 Overall, theresults support the view that stronger land property rights, in the form ofland titles, increase agricultural investment by households.Table 6.2 investigates whether the effect of LUCs on irrigation differs across

regions. The table repeats regression (1) in Table 6.1 for each of the five regionsanalysed. The results are striking. There are no significant effects of LUCs inthe deltas and the Central Coast. In the North and the Central Highlands, onthe other hand, the effect of land titles is strong and highly statisticallysignificant. Plots are 9–12 percentage points more likely to be irrigated aftertitling than before. Within the North, I conducted the analyses separately forPhu Tho and for the highland provinces (Dien Bien, Lai Chau, and Lao Cai).As explained in Section 6.5, Phu Tho is mostly a lowland province with amuch higher prevalence of titling than in the other VARHS provinces in thisregion. Results are in line with those in Table 6.2: there is no effect of titling inPhu Tho and a strong significant effect in the other three provinces. Compare

9 There is a positive but small and insignificant effect of LUCs on irrigation from wells.

Table 6.2 Property rights and agricultural investment, region-specific regressions

Dependent variable: Plot irrigated

Red RiverDelta

North CentralCoast

CentralHighlands

MekongRiver Delta

LUC 0.027 0.115*** 0.006 0.094*** 0.096(0.024) (0.036) (0.036) (0.029) (0.068)

Crop choice restricted 0.097*** 0.125*** 0.176*** 0.122** 0.002(0.023) (0.023) (0.023) (0.046) (0.023)

Working-age HH members, log 0.038** �0.017 0.060* 0.137** 0.140**(0.017) (0.030) (0.036) (0.063) (0.058)

Plot fixed effects Yes Yes Yes Yes YesYear fixed effects Yes Yes Yes Yes YesObservations 2,392 3,755 2,310 946 780

Notes: Level of analysis: Plot. Linear probability models. Standard errors in brackets. Standard errors adjusted forcommune level clustering. Only plots with constant area included. *** p<0.01, ** p<0.05, * p<0.1.



134

this with Figure 6.8 (panel a) and the discussion in Section 6.5, which showedthat LUCs are much less prevalent in the highlands than in the lowlands.

In other words, titling matters exactly where it is least common. This providesa strong case for expanding titling programmes in the uplands. Of course,concerns for equity only make the case stronger: the uplands are poorer thanthe lowlands, and the result presented here (combined with those in Newman,Tarp, and van den Broeck 2015) suggest that titling is a way to increase prod-uctivity and therefore household income.

What explains the interregional variation in the effect of LUCs? One mightsuspect that the absence of a statistically significant effect in the lowlands isdue to lack of variation on the dependent variable, that is, that almost all plotsare already irrigated. This is not the case. Even in the lowlands only around80 per cent of plots are irrigated. A more likely reason is that property rightsare more contested in the hills, implying that the protection offered by titlesis more important. For example, many plots in the uplands are acquired byclearing forest land, which in many cases is communally owned. Hence,disputes about rightful ownership may easily arise. In contrast, land clearingis almost entirely absent in the lowlands.

6.7 Conclusions

This chapter investigated a number of topics related to agricultural land. First,I showed that landlessness among the VARHS panel households is low(around 8 per cent) and stable. Landlessness is highest in the richest incomequintile and lowest in the poorest quintile, supporting the view advanced byRavallion and van de Walle (2008) that Vietnamese households typically donot become landless as a result of adverse, economic shocks, but rather as partof a strategy aimed at exploiting new opportunities in the non-farm economy.Second, I showed that the median farm size is small (around one-third of ahectare—one-fifth of a hectare in the Red River Delta), with a slight declineover time. Hence, there is no evidence of ‘inter-farm’ land consolidation (thatis, no evidence of small farms being merged into larger ones). On the otherhand, I found some evidence that ‘intra-farm’ consolidation is taking place.The mean number of plots operated dropped from 5.8 in 2006 to 4.1 in 2014,and there was a moderate increase in median plot size. This is consistent withthe view that land consolidation programmes are, to some extent, effective interms of merging small land plots into larger ones. Plots remain very small,though (the median plot is 625 m2, one-sixteenth of a hectare).

The chapter also considered land markets. I showed that land sales marketsare more active in the Central Highlands than in any other region by orders ofmagnitude. The likely reason is the high level of migration, combined with

Land Issues

135

the rapid changes in economic circumstances, related, for example, to thecoffee boom, in these provinces. Rich households are more active than poorhouseholds on the demand as well as the supply side of the land salesmarket. Hence, these markets mainly serve the better-off part of the popu-lation. Land rental markets are different. On the supply side, the richerhouseholds are much more likely to participate than poorer households.On the demand side, there is no such correlation. This implies that landrental markets transfer land from rich to poor households. The interregionaldistribution of rental market participation is also very different from thedistribution of sales market activity. Rental markets are most active in theRed River Delta and least active in the North and the Central Highlands.While activity levels in land sales markets are approximately constant overtime, participation rates in rental markets appear to be increasing. The shareof households involved in land rental markets increased from 28 per cent in2006 to 34 per cent in 2014.Finally, I investigated property rights to agricultural land. I found that about

80 per cent of plots are heldwith an LUC (referred to here as a title) and that thisshare is approximately constant over the period of study. There is substantial,interregional variation in land titling. In the Northern Uplands (North exclud-ing Phu Tho province), about 45 per cent of plots are not titled, while thecorresponding share in the MRD is only about 2 per cent. Richer householdsare significantly more likely to hold titled plots than poorer households.I used fixed effects regressions at the plot level to investigate the effects of

LUCson agricultural investment.While I foundno evidence that LUCs increaseinvestment in perennial crops, results indicate that LUCs have a substantial,positive effect on private household investment in irrigation. Remarkably, thiseffect is much stronger in the uplands than in the lowlands. This is paradoxicalbecause titling efforts have been disproportionately focused on the lowlands.These findings provide a strong rationale for expanding systematic land-titlingprogrammes in the Northern Uplands and Central Highlands.

Acknowledgements

I am grateful for inputs from Carol Newman, Finn Tarp, Ulrik Beck, and seminarparticipants in Hanoi.

References

Anderson, J. and S. Davidsen (2011). Recognizing and Reducing Corruption Risks in LandManagement in Viet Nam. Hanoi: National Political Publishing House (Su That).


136

Besley, T. (1995). ‘Property Rights and Investment Incentives: Theory and Evidencefrom Ghana’. Journal of Political Economy, 103(5): 903–37.

Brandt, L. (2006). ‘Land Access, Land Markets and their Distributional Implications inRural Viet Nam’. Summary report. Mimeo. University of Toronto.

Braselle, A. S., F. Gaspart, and J.-P. Platteau (2002). ‘Land Tenure Security and InvestmentIncentives: Puzzling Evidence fromBurkina Faso’. Journal of Development Economics, 67:373–418.

Carter, M. and P. Olinto (2003). ‘Getting Institutions ‘Right’ for Whom? Credit Con-straints and the Impact of Property Rights on the Quantity and Composition ofInvestment’. American Journal of Agricultural Economics, 85(1): 173–86.

Deininger, K. and S. Jin (2008). ‘Land Sales and Rental Markets in Transition: Evidencefrom Rural Viet Nam’. Oxford Bulletin of Economics and Statistics, 70(1): 67–101.

Do, Q.T. and L. Iyer (2008). ‘Land-Titling and Rural Transition in Viet Nam’. EconomicDevelopment and Cultural Change, 56(3): 531–79.

Feder, G. and T. Onchan (1987). ‘Land Ownership Security and Farm Investment inThailand’. American Journal of Agricultural Economics, 69: 311–20.

Gourou, P. (1936). Les Paysans du Delta Tonkinois. Paris: Mouton &Co., andMaison DesScience de L’Homme.

Hornbeck, R. (2010). ‘Barbed Wire: Property Rights and Agricultural Development’.Quarterly Journal of Economics, 125(2): 767–810.

Jacoby, H. and G. Mansuri (2008). ‘Land Tenancy and Non-Contractible Investment inRural Pakistan’. Review of Economic Studies, 73(3): 763–88.

Kerkvliet, B. T. (2006). ‘Agricultural Land in Viet Nam: Markets Tempered by Family,Community and Socialist Practices’. Journal of Agrarian Change, 6(3): 285–305.

Khai, L. D., T. Markussen, S. McCoy, and F. Tarp (2013). ‘Access to Land: Market- andNon-Market Land Transactions in Rural Viet Nam’. In S. Holden., K. Otsuka, andK. Deininger (eds), Land Tenure Reform in Asia and Africa: Assessing Impacts on Povertyand Natural Resource Management. Basingstoke: Palgrave Macmillan.

Markussen, T. (2008). ‘Property Rights, Productivity and Common Property Resources:Insights from Rural Cambodia’. World Development, 36(11): 2277–96.

Markussen, T. and F. Tarp (2014). ‘Political Connections and Land-Related Investmentin Rural Viet Nam’. Journal of Development Economics, 110: 191–302.

Markussen, T., F. Tarp, and K. van den Broeck (2011). ‘The Forgotten Property Rights:Evidence on Land Use Rights in Viet Nam’. World Development, 39(5): 839–50.

Markussen, T., D. H. Thiep, N. D. A. Tuan, and F. Tarp (2013). ‘Inter- and Intra-FarmLand Fragmentation in Viet Nam’. CIEMWorking Paper. Hanoi: Central Institute forEconomic Management.

Marsh, S. P., T. G. MacAulay, and P. V. Hung (2006). Agricultural Development and LandPolicy in Vietnam. Canberra: ACIAR.

Newman, C., F. Tarp, and K. van den Broeck (2015). ‘Property Rights and Productivity:The Case of Joint Land-Titling in Viet Nam’. Land Economics, 91(1): 91–105.

Pingali, P. and V. T. Xuan (1992). ‘Viet Nam: Decollectivization and Rice ProductivityGrowth’. Economic Development and Cultural Change, 40(4): 697–718.

Popkin, S. L. (1979). The Rational Peasant: The Political Economy of Rural Society in VietNam. Berkeley, CA: University of California Press.

Land Issues

137

Ravallion, M. and D. van de Walle (2004). ‘Breaking Up the Collective Farms: WelfareOutcomes of Viet Nam’s Massive Land Privatization’. Economics of Transition, 12:201–36.

Ravallion, M. and D. van deWalle (2006). ‘Land Reallocation in an Agrarian Transition’.Economic Journal, 116: 924–42.

Ravallion, M. and D. van de Walle (2008). ‘Does Rising Landlessness Signal Successor Failure for Viet Nam’s Agrarian Transition?’ Journal of Development Economics, 87:191–209.

Rozelle, S. and J. F. M. Swinnen (2004). ‘Success and Failure of Reform: Insights from theTransition of Agriculture’. Journal of Economic Literature, 42: 404–56.


138

7

Labour and Migration

Gaia Narciso

7.1 Introduction

According to the 2009 Vietnamese census, 6.6 million people migrated withinViet Nam over the period 2004–9 (United Nations Vietnam 2010), an increaseof 46 per cent with respect to the number of internal migrants recorded in the1999 census. The 2004 Viet Nam Household Living Standard Survey (VHLSS)unveils that almost 89 per cent of households with a migrant receive remit-tances (United Nations Vietnam 2010), which constitute a substantial meansby which households can pay daily expenses such as education or health-careexpenses.

The aim of this chapter is to provide an overview of the characteristics ofsending households and analyse the labour market effects of migration inrural Viet Nam, on the basis of the Viet Nam Access to Resources HouseholdSurvey (VARHS) conducted in 2012 and 2014. The economics literature hasextensively explored the determinants of migration. The seminal article byHarris and Todaro (1970) modelled the rural to urban migration decision.According to their theory, the main determinant of migration is the expectedwage differential between the origin place of residence and the destination.Later contributions to the literature analysed other factors besides wage differ-entials and introduced income uncertainty and relative deprivation as furtherdeterminants of the migration decision (Stark 1991). The new economics ofmigrationmodelled themigration decision as a risk-sharing decision, wherebyhouseholds can diversify risk by letting a member migrate to another labourmarket, with the aim of reducing the income risk facing households.1

1 See Bauer and Zimmermann (1994) for an extensive review of the literature.

This chapter discusses differences across households with a migrant on thebasis of reasons for migrating and explores the features of migrants andsending households. We try to establish whether a positive or negative self-selection of migrants can be identified. In particular, we focus on the labourmarket effects of migration. We investigate the move out of agriculture intomore waged employment in urban and rural areas, thus complementing thefindings presented in Chapter 4 on the non-farm rural economy. Next, weexamine the households that receive remittances and how they are used.Finally, we uncover the role of migration as shock-coping mechanisms inrural Viet Nam.This chapter is organized as follows. Section 7.2 provides a policy back-

ground on migration directives in Viet Nam and an overview of the literature.Section 7.3 describes the data, while Section 7.4 investigates the features ofsending households. Section 7.5 discusses the characteristics of migrants,while remittance behaviour is explored in Section 7.6. Section 7.7 presentsthe results of the econometric investigation of the role of migration as a risk-coping mechanism, while Section 7.8 investigates the relationship betweenmigration and access to credit. Section 7.9 concludes.

7.2 Policy Background and Literature Review

The ‘Doi Moi’ policy, introduced in Viet Nam in 1986, led to a drastic increasein domestic migration, in response to the rapid economic growth experiencedwith the opening up of the economy. Moreover, since 1986, Viet Nam hasseen an increase in the population leading to a shortage of arable land in thecountryside. This has motivated many individuals to move from rural to urbanareas, where industrial development offers more employment opportunities.Census 2009 figures for ‘unplanned’ internal migration in Viet Nam reveal

that migration between provinces reached 1.3 million individuals, about 2.5per cent of the total population, in 1989, 2 million or 2.9 per cent of the totalpopulation in 1999, and 3.4 million or 4.3 per cent of the total population in2009. Furthermore, the annual rate of migration within provinces increasedfrom 0.6 per cent in 1999 to 4.2 per cent in 2009. Forecasts predict thatmigration will continue to rise, reaching 6 million or 6.4 per cent of thetotal population by 2019 (GSO 2011).The socioeconomic repercussions of migration have spurred the Viet

Nam government to implement a number of national regulations aimed atmanaging internal migration. The range of Decrees and Decisions aim toensure socioeconomic development, political security and social safety inthe receiving areas. The Decisions also state migrants’ responsibilities.


140

In particular, migrants are expected to fully comply with the regulations onmigration, in relation to civil registration (the ho khau system).2

A few studies have investigated patterns ofmigration in Viet Nam. Using theVHLSS, Nguyen and co-authors (2008) explore the determinants of migrationin Viet Nam. The authors provide evidence that larger households and house-holds with a higher level of education tend to be associated with higheremigration rates. Moreover, households involved in waged employment aremore likely to migrate. A recent work by Nguyen, Raabe, and Grote (2015)explores the relationship between shocks and rural–urban migration. Theauthors provide evidence that migration acts as a risk-coping mechanism.Gröger and Zylberberg (2016) analyse in particular the effect of a typhoon,which hit central Viet Nam in 2009. Internal labour migration could beregarded as being a shock-coping strategy in rural economies when house-holds cannot rely on remittances. Indeed, the analysis predicts that, after atyphoon, family members are more likely to migrate and support their rela-tives through remittances.

At a more macro level, Phan and Coxhead (2010) explore the determinantsof inter-provincial migration and the effect of migration on inter-provincialinequality. Using a gravity model, the authors show that migrants move fromlow-income to high-income provinces. As for the impact of migration oninequality, the evidence suggests that on average migration leads to a reduc-tion in inequality, although the extent of the effect mainly depends on thetype of receiving province.

We contribute to the existing literature by providing more recent evidenceon the role of internal migration in rural Viet Nam.

7.3 Data

Our data come from the 2012 and 2014 VARHS, which provide a detailedpicture of the incomes, assets, and access to resources of rural households intwelve provinces.3 While data have been gathered using this survey instru-ment since 2006, in 2012, a new module was introduced to capture informa-tion on migration.4

2 It is estimated that over 5 million Vietnamese do not have permanent registration where theylive and are therefore excluded from health-care provision, access to schooling, and socialprotection (World Bank 2016). Loosening the link between access to services and registrationwould enhance equality of opportunities for migrants (World Bank 2016).

3 See CIEM (2011) and CIEM, DERG, and IPSARD (2013) for comprehensive descriptive reportsof the data gathered in each round of the survey.

4 The analysis in this chapter relies on the information provided by sending households only.


141

According to the 2012 VARHS, about 20 per cent of interviewed householdshave at least one member who has migrated, of which 48 per cent are workingmigrants. We do not observe much variation over time, as in 2014 the percent-ages of households with a migrant and households with a working migrant areindeed very similar (19.60 and 48 per cent, respectively). About 22 per cent ofsending households have a permanent migrant in 2012, while 63 per cent ofsending households have a migrant who is only away temporarily. Two yearslater, about 15 per cent of sending households have at least one permanentmigrant, while 69 per cent have at least one temporary migrant.The majority of migration occurs across provinces. In 2012, about 62 per

cent of the sending households reported that the migrant migrated outside ofthe province of origin, while 37 per cent of migrants moved within theprovince. Less than 1 per cent moved internationally. Working migrants areless likely to move within the province of origin and are more likely either tomove to another province or to move internationally (see Table 7.1). Weobserve a significant increase in inter-province migration in 2014, as 73 percent of migrants moved to another province. A significant increase is alsonoted in international migration, as 10 per cent of working migrants arereported to have migrated abroad.Table 7.2 presents the percentage of households with a migrant by province

and the percentage of households with a working migrant. According to the2012 VARHS, the province with the highest percentage of migrant householdsis Nghe An, where about 47 per cent of interviewed households have at leastone migrant living away, while about 36 per cent of households have a workingmigrant. Quang Nam also reports a high percentage of households with amigrant (27 per cent), although it shows a smaller fraction of households witha working migrant (8.8 per cent). The data from the 2014 survey show someinteresting changes in the percentages of migrant households by province.Three provinces in particular, Dak Lak, Dak Nong, and Lam Dong, report highpercentages of migrant households, around 28 per cent. Most of the provincesshow a remarkable increase in the number of households with a migrant.It appears indeed that migration is continuing to rise at a remarkable speed.

Table 7.1 Inter-province and intra-province migration

2012 2014

All migrants (%) Working migrants (%) All migrants (%) Working migrants (%)

Same province 37.55 34.06 20.06 15.30Another province 61.90 65 73.30 74.14Abroad 0.55 0.94 6.64 10.55

Source: Author’s calculations based on the VARHS database.


142

Where domigrants move to? Table 7.3 reports the list of the main provincesreceiving migrants. Ha Noi and Ho Chi Minh provinces received the highestshare of migrants in our sample in 2012, 26.55 and 16.51 per cent, respect-ively, supporting the idea that migrants tend to converge in big urban cities.5

This pattern is even more remarkable in 2014, as Ha Noi and Ho Chi Minhprovinces attracted 26.99 and 20.55 per cent share of migrants, respectively,in our sample.

Table 7.4 presents the reasons for migration, distinguishing between tem-porary and permanentmigrants. Themajority of temporarymigrants are away

Table 7.2 Province of origin

Province 2012 2014

Households witha migrant (%)

Households with aworking migrant (%)

Households witha migrant (%)

Households with aworking migrant (%)

Ha Tay 18.02 9.18 17.32 9.34Lao Cai 18.09 9.52 5.61 3.74Phu Tho 16.71 6.23 20.78 10.65Lai Chau 7.69 1.54 15.55 5.18Dien Biem 14.29 7.32 24.41 7.09Nghe An 47.11 36.44 24.12 16.67Quang Nam 27.22 8.88 17.45 7.99Khanh Hoa 17.71 7.29 26.85 17.59Dak Lak 17.68 7.31 28.39 8.02Dak Nong 17.70 7.96 28.15 11.85Lam Dong 20.78 2.60 28.20 8.97Long An 7.19 3.12 13.51 6.61


Table 7.3 Province of destination

2012 2014

Observations % Observations %

Ha Noi 193 26.55 176 26.99Ho Chi Minh 120 16.51 134 20.55Da Nang 70 9.63 49 7.52Nghe An 40 5.50 19 2.91Quang Nam 37 5.09 7 1.07Binh Duong 24 3.30 14 2.15Phu Tho 22 3.03 15 2.30Dien Bien 21 2.89 22 3.37Dak Lak 19 2.61 26 3.99


5 These results might be driven by the distribution of provinces in our sample.


143

due to education and work, while the majority of permanent migrants areaway either for family reunification or for work reasons. Army service alsoplays a role, with about 4 per cent of migrants away on army duty.

7.4 Household Characteristics

Are households with a migrant wealthier? In order to address this issue weconsider the distribution of sending and non-sending households by expend-iture quintile. The results are shown in Table 7.5. A smaller percentage ofsending households is in the first food expenditure quintile.6 The differenceis particularly striking if we look at households with a working migrant, wherethe percentage of households in the first quintile in 2012 is just 11.34 per centcompared to 22.28 per cent of households with no migrants. A much higherpercentage of households with a working migrant is in the last food expend-iture quintile, therefore indicating that households with a working migrantare wealthier. The distribution of sending and non-sending householdsappears to be unchanged in 2014. The aim of Table 7.5 is to present a simplebut informative correlation between household wealth and migration status.However, we cannot infer from these summary statistics whether sendinghouseholds are wealthier because they have a migrant away (and potentiallyreceive remittances) or whether they were able to send amigrant away becausethey are wealthier. Also, workingmigrants are likely to be wealthier than othermigrants, as they are more likely to be educated and therefore better off.

Table 7.4 Reasons for migrating

All migrants (%) Temporary migrants (%) Permanent migrants (%)

2012Work/looking for work 45.29 46.05 40Education 35.60 46.49 1.29Marriage/family reunification 13.62 1.1 52.26Army service 3.80 5.26 1.94Other 1.08 0.8 4.51

2014Work/looking for work 45.54 47.06 24.76Education 36.63 44.57 1.90Marriage/family reunification 10.72 2.27 60Army service 4.04 4.75 0.95Other 3.06 1.36 12.38


6 This result is in line with the findings presented in Chapter 10 on welfare dynamics.


144

Table 7.6 compares sending and non-sending households in terms of a set ofdemographic features. Heads of non-sending households tend to be olderthan sending household heads and the difference is statistically significantat the 5 per cent level in 2014, while no statistically significant differenceappears in 2012. Sending households have a higher net income than non-sending households and the difference is statistically significant in both years.This finding is indeed consistent with the summary statistics presented inTable 7.5 on food expenditure quintiles. Ethnicity also seems to play a role.A higher percentage of households with a migrant belong to the Kinh ethnic

Table 7.5 Distribution of households by migration status and food expenditure quintile

Food expenditurequintile

Distribution ofhouseholds with amigrant (%)

Distribution ofhouseholds with aworking migrant (%)

Distribution ofhouseholds withno migrant (%)

20121 13.31 11.34 22.282 18.40 18.62 21.183 19.96 23.48 18.614 19.37 16.60 20.185 28.96 29.96 17.75

20141 14.42 10.85 20.772 14.42 12.02 20.683 21.35 21.71 20.954 20.60 21.32 19.855 29.21 34.11 17.75


Table 7.6 Household characteristics by migration status

Variable Households with a migrant Households with no migrant Difference

(1) (2) (1)–(2)

2012Age 41.59 41.61 �0.02Net income (’000 VND) 2270 1820 450***Kinh 87.47% 78.16% 0.09***Economic shock 18.98% 18.51% 0.00Natural shock 37.38% 28.26% 0.09***

2014Age 39.84 43.82 �3.97***Net income (’000 VND) 2467 1940 527***Kinh 82.21% 79.05% 0.03Economic shock 13.67% 13.19% 0.01Natural shock 25.47% 22.73% 0.03

Note: * significant at 10%; ** significant at 5%; *** significant at 1%.



145

group in 2012, compared to non-sending households, suggesting that theyeither have more opportunities for migration or are more willing to do so.7

Finally, a larger proportion of sending households are affected by naturalshocks in 2012, but no difference appears to exist in terms of exposure toshocks in 2014.Given the different reasons for migrating, Table 7.7 presents the character-

istics of households with a working migrant with respect to households withno non-working migrant. Heads of households with a working migrant tendto be older than non-working migrant household heads and the difference isstatistically significant in 2014. Net household income is higher in house-holds with a working migrant in both years. Kinh households are found to bemore likely to have a working migrant, although this difference is statisticallysignificant in 2014 only. Regarding exposure to shocks, we do not find muchdifference between households with a working migrant with respect to house-holds with no working migrant in either year. We explore this aspect in theregression analysis in Section 7.7.

Table 7.7 Household characteristics by reason of migration

Variable Households with aworking migrant

Households withother migrant

Difference

(1) (2) (1)�(2)

2012Age of household head 42.75 40.51 2.24Net income (’000 VND) 2534 2024 510*Kinh 89.47% 85.60% 0.04Total land owned (ha) 5034.41 7194.00 �2159.59**Economic shock 16.60% 21.21% �0.05Natural shock 40.89% 34.09% 0.07

2014Age of household head 41.60 38.21 3.58***Net income (’000 VND) 2732 2220 511***Kinh 86.43% 78.26% 0.08**Total land owned (ha) 6503.15 6772.14 268.99Economic shock 12.79% 14.49% �0.02Natural shock 24.80% 26.09% �0.01

Note:* significant at 10%; ** significant at 5%; *** significant at 1%.


7 According to the findings in Chapter 4, ethnic minorities are more likely to transition out ofspecialized agriculture; i.e. are more likely to diversify activities. It is interesting to note that such adiversification does not include location mobility.


146

7.5 Migrant Characteristics

Table 7.8 presents the characteristics of migrants by comparing workingmigrants with non-working migrants. A slight majority of migrants are men,although the percentage is higher for working migrants in both years. About30 per cent of migrants aremarried, while this percentage slightly increases forworking migrants. Working migrants tend to leave the commune later thanother types of migrants, which might be related to the fact that they are morelikely to receive their education before migrating compared to householdsthat migrate to attend school. Indeed a lower percentage of working migrantshave no diploma. There is no difference in the length of the migrationexperience between the two groups. On average, migrants have been awayfor two years. There does not seem to be any statistically significant differencebetween working and non-working migrants in terms of the intended lengthof stay in 2012, although this difference becomes statistically significant in2014: it appears that working migrants are more likely to return to their homecommunity. This result is not unexpected, given thatmigrants whomoved forfamily reasons are less likely to return to their home communities.

What do migrants do? In the light of labour market movements, it iscrucial to understand what migrants’ occupations are during their migrationexperience. Table 7.9 presents the percentage of working migrants by occupa-tion. The majority of migrants are employed in manual jobs and they workeither as unskilled workers or as skilled workers. A significant percentage of

Table 7.8 Migrant and working migrant characteristics

Migrants characteristics (variable) All migrants Working migrants t-Test ofdifference

Mean SD Mean SD

2012Male 51.05% 0.50 58.96% 0.49 ***Married 30.50% 0.46 36.70% 0.48 ***Age at migration 22.45 8.06 25.39 9.14 ***No diploma 62.43% 48.46 40.46% 0.49 ***Years since the migrant left 2.14 1.95 2.05 2.01Permanent 25.37% 0.43 22.79% 0.42

2014Male 52.78% 0.50 57.29% 0.49 ***Married 27.99% 0.45 32.22% 0.47 ***Age at migration 22.62 8.16 24.50 8.86 ***No diploma 63.65% 0.48 47.83% 0.50 ***Years since the migrant left 2.07 1.90 2.13 2.13Permanent 19.19% 0.39 13.78 0.34 ***




147

migrant workers are employed in top or mid-level occupations. With respectto 2012, the 2014 data show a decrease in the percentage of migrants involvedin unskilled occupations. At the same time, migrants appear to be moreinvolved in mid-level occupations, across all fields. The summary statisticspresented in Table 7.9 further support the evidence shown in Chapter 4regarding the movement out of agricultural activities in Viet Nam.Given the level of inter-province migration, it is also interesting to explore

how migrants manage to find their job at the destination. The literature onmigration networks explores the role of family and friends in providinginformation about job opportunities to potential or recent migrants. Interest-ingly, in the case of Viet Nam, the role of migration networks in providingsupport to migrants seems more limited. Table 7.10 presents the evidence.About one-third of migrants in the sample found a job through their migra-tion network (i.e. family and friends). However, the majority found an occu-pation in the location of destination either through an employment serviceor, more generally, through self-seeking. This is a rather interesting patternthat suggests migrants may have migrated to a specific destination withoutthe support of an existing migration network.

Table 7.9 Migrant occupation

2012 (%) 2014 (%)

Army 3.96 1.74Leaders in all fields and levels 7.25 2.48Top-level occupations in all fields 7.25 9.93Mid-level occupations in all fields 5.71 20.60Staff (elementary occupations, white-collar technical personnel) 9.45 4.96Skilled workers in personal services, security protection, and sales 2.86 5.96Skilled workers in agriculture, forestry, and aquaculture 1.54 0.25Skilled handicraftsmen and other related skilled manual workers 19.78 17.87Assemblers and machine operators 7.69 8.93Unskilled workers 33.41 26.55Communal officials who are not public servants 0.88 0.74


Table 7.10 Role of migration networks

How did the migrant get the job? 2012 (%) 2014 (%)

Self-seeking 57.45 51.77Relative/friend 30.50 34.09Employment service 4.96 5.81Other 7.09 8.34



148

7.6 Remittance Behaviour

Migrants may send remittances for altruistic motives, a sense of social respon-sibility; as a risk-sharing mechanism, to smooth consumption in the face ofexternal shocks; or as a combination of these reasons (Maimbo and Ratha2005). Although our data do not allow us to uncover the motives for sendingremittances, we can explore the characteristics of those who receive remit-tances and those who do not and analyse the reasons for sending as reportedby the receiving households. We observe a slight increase in the percentage ofhouseholds receiving remittances: about 26 per cent of migrant households inour sample received remittances in 2012, while the percentage rose to 30 percent in 2014.8 Remittance-recipient households differ on many aspects withrespect to migrant households that do not receive remittances. Table 7.11shows that remittance-recipient households have an older household headthan non-remittance-recipient households in either year. Net householdincome of remittance-recipient households is greater and the difference isstatistically significant at the 1 per cent level in 2014. We will further inves-tigate the role of remittances in boosting household welfare in the nextsection. Ethnicity appears to matter, as Kinh households tend to receivemore remittances than other ethnic minorities household—the difference isstatistically significant at the 5 per cent level in 2014. Finally, remittance-recipient households are as likely as non-remittance-recipient households tobe affected by economic of shocks in either year, although they are more likely

Table 7.11 Remittance-recipient and non-remittance-recipient household characteristics

Variable Remittance-recipienthouseholds (1)

Non-remittance-recipienthouseholds (2)

Difference(1)�(2)

2012Age of household head 46.20 40.02 6.18***Net income (’000 VND) 2602 2158 444Kinh 91.54% 86.09% 0.05Economic shock 14.61% 20.47% �0.06Natural shock 48.46% 33.60% 0.15***

2014Age of household head 42.40 38.72 3.69***Net income (’000 VND) 2904 2273 630 ***Kinh 87.19% 80% 0.07**Economic shock 14.02% 13.51% 0.00Natural shock 25.61% 25.40% 0.00



8 Chapter 9 further investigates the role of private transfers in the household incomecomposition.


149

to experience natural shocks in 2012. We explore further the relationshipbetween remittances and shocks in the next section.A recent strand of the migration literature has focused on the ability of

migrants to control how remittances are used. The issue is relevant given theasymmetric information that characterizes the relationship between migrantsand their family of origin. Ashraf and colleagues (2015), Batista and Narciso(forthcoming), Elsner, Narciso, and Thijssen (2013), and McKenzie, Gibson,and Stillman (2013) show that spatial distance and lack of monitoring harmsthe quality of information flows betweenmigrants and their family and friendsin the commune of origin. Table 7.12 compares how remittances are used byhouseholds, with respect to migrants’ purpose for sending remittances.According to column 1, remittances are mainly spent for daily expenses (i.e.

daily consumption and bills). The second largest category is savings, followedby expenses for special occasions and medical and educational expenses.9

There is no statistically significant difference between the way householdsspend the remittances and the migrants’ purpose of sending remittances. Thisfinding differs with respect to previous results found in the literature, but it islikely to be driven by the fact that the remittance-recipients have a biased viewof what the migrant’s purpose for sending remittances is and might simplyrespond to the question in a way that validates the way they spend theremittances.There is some evidence that migrants receive transfers from the household

of origin as well. About a third of all migrants in our sample receive transfers,

Table 7.12 Remittance use

How household spendsremittances (%)

Migrant’s purpose forsending remittances (%)

2012Daily meals and bills 44.57 46.86Medical expenses 6.86 5.14Educational expenses 5.14 5.71Savings 14.29 14.86Special occasion 6.86 6.86House 9.14 7.43

2014Daily meals and bills 56.72 55.72Medical expenses 6.47 7.46Educational expenses 5.47 5.47Savings 11.44 13.43Special occasion 1.49 1.49House 2.99 2.49


9 The majority of migrants send remittances every month.


150

a result which is mainly driven by the large number of migrants who movedformotives of education. However, it is interesting to note that a percentage ofworking migrants also receive transfers (7 per cent in 2012, 15 per cent in2014), therefore highlighting the potential vulnerability working migrantsface—an issue that needs further investigation in future research.

7.7 How Does Migration Impact the Welfareof Sending Households?

To explore this question we create a household panel that tracks sending andnon-sending households in 2012 and 2014. We consider the extent to whichmigration serves as a risk-copingmechanism and estimate the followingmodel:

ΔFoodExp pcht ¼ β1migrantht þ β2shockht þX0htγþ αh þ τt þ εht; ð1Þ

where ΔFoodExp_pcht is the change in household food expenditure per capita,for household h at time t; the variablemigrantht takes the value 1 if householdh has a migrant away at time t and 0 otherwise; the indicator variable shockhtmeasures whether the household experience a shock (either economic ornatural shock); and Xht is a vector of household characteristics, such as ethni-city, an indicator variable for remittance-recipient households, age of thehousehold head, and whether the household head is a woman. We alsoinclude household fixed effects (αh) and time fixed effects (τt). Table 7.13presents the results of this simple exercise.

The estimated coefficient of the shock variable is not statistically significant(Table 7.13, column 1), a result which is likely due to the different nature ofshocks a household can face. Interestingly, migrant households show higherfood expenditure per capita and the relationship is statistically significant atthe 1 per cent level. This result holds also when we control for the set ofhousehold characteristics (column 2). In column 3 we interact the shockdummy variable with the indicator variable of being a sending household.We find that sending households are not affected by shocks differently fromnon-sending households. Of course, the reason for migrating is very relevant;therefore, in the next column we distinguish between working migrants andmigrants who left the household for other reasons such as education, familyreunification, or military service. Column 4 shows that having a migrantoutside the household has a positive and statistically significant impact onthe change in per capita food expenditure, for households with a workingmigrant and households with another type of migrant, relative to non-sending households. The results hold also when we control for other house-hold characteristics, such as age of the household head, ethnicity, andwhether the household head is a woman (column 5). Finally, in column 6,


151

we interact the shock dummy variable with the indicator variable of having amigrant, distinguishing between working migrants and other migrants. Wefind that the coefficient of the interaction term is not statistically significant,while the relation between both type of migrant households and the changein per capita food expenditure are still positive and statistically significant.The variable shock captures both economic and natural shocks. Given the

potential endogeneity between economic shocks and household behaviour,we repeat the previous analysis and focus on natural shocks only.Table 7.14 analyses the impact of migration and natural shocks on the

change in food expenditure. Again, migration is associated with a positiveand statistically significant increase in food expenditure, while the estimatedcoefficient on natural shocks is negative but it is not statistically significant.These findings hold also when we control for household characteristics(column 2). Next, we interact the migrant household dummy variable withthe natural shock indicator. Migration seems to act as a natural shock-copingmechanism as sending households are able to offset the impact of the naturalshock on the change in per capita food expenditure. In columns 4–6,

Table 7.13 Migration and food expenditure

Variables Change in per capita food expenditure

(1) (2) (3) (4) (5) (6)

Shock 4.69 4.32 �0.76 4.66 4.28 �0.75(16.505) (16.581) (17.840) (16.535) (16.611) (17.848)

Migrant 85.49*** 84.05*** 75.24***(20.331) (20.418) (27.263)

Migrant � shock 20.85(38.517)

Kinh �18.09 �19.27 �18.14 �22.31(189.883) (190.547) (190.025) (189.797)

Age of household head 0.83 0.82 0.83 0.82(0.794) (0.790) (0.793) (0.792)

Female household head 44.80 44.91 44.80 45.08(58.360) (58.386) (58.352) (58.379)

Working migrant 87.81*** 86.10*** 67.92*(28.502) (28.574) (39.651)

Other migrant 83.42*** 82.23*** 81.14**(25.542) (25.599) (33.062)

Working migrant � shock 43.52(53.394)

Other migrant � shock 2.36(48.656)

Observations 4,739 4,738 4,738 4,739 4,738 4,738Number of households 0.024 0.025 0.025 0.024 0.025 0.025Adjusted R-squared 2,716 2,715 2,715 2,716 2,715 2,715

Notes: Each model includes household and time fixed effects. Robust standard errors clustered at the household level inparentheses. * significant at 10%; ** significant at 5%; *** significant at 1%.



152

we distinguish between the reasons for migrating. Working migrants are posi-tively associated with a change in food expenditure and so are other types ofmigrants. A word of caution is needed here. Wealthier households are morelikely to send their children to study away from home (other migrant). Thiscould explain the positive and statistically significant coefficient on the othermigrant variable. On the other hand, having a working migrant might signalthat the household is less wealthy and therefore had to send a member to worksomewhere else. Interestingly, having a working migrant offsets the impactof negative shocks on the change in food expenditure (column 6). Again, thisresult highlights the importance of migration as a shock-coping mechanism.

Table 7.15 explores the role of remittances in acting as a coping mechanismin the event of negative shocks. We replicate our analysis by focusing onremittance-recipient households, rather than just households with a migrantaway. The estimation results in column 1 show that being a remittance-recipient household is positively correlated with food expenditure. On aver-age, food expenditure is greater in remittance-recipient households: thisresult highlights the importance of remittances as a means of covering food

Table 7.14 Migration and natural shocks


(1) (2) (3) (4) (5) (6)

Natural shock �15.55 �15.95 �36.88* �15.58 �15.97 �36.43*(17.781) (17.743) (19.160) (17.789) (17.752) (19.142)

Migrant 85.92*** 84.47*** 57.35**(20.330) (20.417) (26.289)

Migrant � natural 81.22**shock (38.820)Kinh �21.45 �16.57 �21.50 �16.16

(189.400) (191.925) (189.557) (191.410)Age of household head 0.84 0.86 0.83 0.85

(0.807) (0.806) (0.807) (0.812)Female household head 45.62 45.96 45.61 45.22

(58.226) (58.020) (58.220) (57.948)Working migrant 88.49*** 86.76*** 45.68

(28.479) (28.546) (37.095)Other migrant 83.63*** 82.44*** 67.59**

(25.550) (25.611) (32.681)Working migrant �

natural shock123.15**(53.025)

Other migrant �natural shock

44.01(49.768)

Observations 4,739 4,738 4,738 4,739 4,738 4,738Number of households 0.025 0.025 0.027 0.025 0.025 0.028Adjusted R-squared 2,716 2,715 2,715 2,716 2,715 2,715

Notes: Each model includes household and time fixed effects. Robust standard errors clustered at the household level inparentheses. * significant at 10%; ** significant at 5%; *** significant at 1%.



153

expenses for the receiving household. We also interact the dummy variablecapturing remittance-recipient household with the shock dummy variable toinvestigate the role of remittances as a risk-coping mechanism.10 The esti-mated coefficient on the interaction term between remittances and shock isnot statistically significant. Therefore, we may conclude that, althoughremittance-recipient household do display higher food expenditure, we donot find robust evidence of remittances as a shock-copingmechanism.Wewillexplore further this aspect in our analysis of the role of remittances in easingaccess to credit in the next section. Similar results hold when we focus ouranalysis on the effect of natural shocks (column 2).

7.8 Migration and Access to Credit

How does migration affect the financial behaviour of households? The evi-dence reported in Table 7.16 shows that households with a working migrantand other migrant households show no statistically significant relationship

Table 7.15 Remittances and food expenditure


(1) (2)

Shock 10.13(16.795)

Remittance-recipient household 189.45*** 151.88***(62.278) (57.558)

Remittance-recipient household � shock �113.81(76.771)

Kinh �3.50 �18.09(197.911) (196.105)

Age of household head 0.95 0.95(0.732) (0.745)

Female household head 42.24 43.01(58.188) (58.256)

Natural shock �13.54(18.120)

Remittance-recipient household � shock �34.40(81.915)

Observations 4,738 4,738Number of households 0.023 0.023Adjusted R-squared 2,715 2,715

Notes: Each model includes household and time fixed effects. Robust standard errors clustered atthe household level in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%.


10 Remittance recipient households are defined as households that receive remittances at leastonce a year.


154

with the change in total amount borrowed. Interestingly, remittance-recipienthouseholds experience an increase in the total amount borrowed, a result thatcan be interpreted as showing that remittances increase collateral and easeaccess to credit. Column 2 presents the results related to the interactionbetween the type of migrant household and natural shocks. We do not finda statistically significant relationship between this interaction and the changein total amount borrowed.

The next column explores the impact of remittances in the presence of naturalshocks. Being a workingmigrant household eases access to credit in the case of anegative natural shock, therefore supporting the view that households with aworking migrant face natural shocks by resorting to more borrowing. On theother hand, remittance-recipient households reduce the amount borrowed inthe case of a negative natural shock. We may conclude that, on the one hand,having a workingmigrant eases access to credit in the case of a natural shock; onthe other, remittances counteract the negative impact of a natural shock byreducing the amount borrowed by the household.

Table 7.16 Migration, remittances, and borrowing behaviour

Variables Change in total amount borrowed (‘000s)

(1) (2) (3)

Natural shock �2.90 �5.82 �5.86(3.103) (4.294) (4.295)

Working migrant �2.98 �6.13 �9.11(5.783) (7.551) (7.936)

Other migrant �4.74 �8.75 �9.05(17.337) (24.770) (24.796)

Working migrant � nat. shock 10.19 20.77**(8.696) (9.821)

Other migrant � nat. shock 12.04 12.66(23.845) (23.892)

Remittance-recipient household 10.41 10.10 19.12**(7.311) (7.409) (9.666)

Remittance-recipient household � �27.17**natural shock (13.740)Kinh 19.92 20.59 19.72

(17.331) (17.658) (17.689)Age of household head �0.14 �0.14 �0.14

(0.112) (0.115) (0.115)Female household head �3.65 �3.63 �3.43

(4.441) (4.427) (4.401)Observations 4,835 4,835 4,835Number of households 0.000 0.001 0.001Adjusted R-squared 2,745 2,745 2,745

Notes: Each model includes household and time fixed effects. Robust standard errors clustered at thehousehold level in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%.



155

7.9 Conclusions

This chapter provides an overview of the characteristics of sending house-holds and analyses the effects of migration in Viet Nam, on the basis of theVARHS conducted in 2012 and 2014. The data reveal significant movementsof household members, both intra-province and inter-province, with about20 per cent of the interviewed households having at least one member whohas migrated. The two main reasons for migrating are education and work-related motives. Significant differences are uncovered between sending andnon-sending households, as households with a migrant are wealthier thannon-sending households, as measured by food expenditure quintiles. Theeconometric analysis shows that migration acts as a shock-coping mechan-ism, especially in the presence of natural shocks. Households with a migrantaway are also more likely to have better access to the market for credit. Inparticular, remittance-recipient households seem to react better to naturalshocks, as the remittance flows counterbalance the need for formal borrowing.Given the large and increasing migration movements within Viet Nam, it has

become crucial to understand the role of migration as a means of poverty reduc-tionandas a risk-copingmechanismandalso the features of sendinghouseholds,especially in the face of shocks affecting household welfare. This chaptermakes a significant first step in understanding these issues for the twelve prov-inces included in the VARHS dataset. The results suggest that migration has thepotential to act as a safety valve for vulnerable households in rural communities.Better-off households are more likely to migrate, however, which suggests thatthere are constraints to migration for less well-off households. Our findingssuggest that constraints to voluntary migration should be removed, particularlyfor poorer households where members may have the desire to leave their homecommunity tofindworkbutmaynothave the resources todo so.Moreover, theremay be a role for government or other agencies in developing formal bankingmechanisms to facilitate the remittance of funds back to sending households.On a final note, we would like to emphasize that the VARHS data focus on thecharacteristics of the sendinghouseholds andnot themigrants themselves.Moredata and research are needed on the vulnerability and welfare of the migrantswhomove to find work. This is beyond the scope of these data and this study.

Acknowledgements

For very helpful comments and discussions, the author thanks Carol Newman, TranThanh Nhan, Finn Tarp, and seminar participants at the Central Institute for EconomicManagement (CIEM) 2015 workshop. The author also thanks Gaspare Tortorici forproviding excellent research assistance.


156

References

Ashraf, N., D. Aycinena, C. Martínez, and D. Yang (2015). ‘Savings in TransnationalHouseholds: A Field Experiment among Migrants from El Salvador’. Review ofEconomics and Statistics, 2(97): 332–51.

Batista, C. and G. Narciso (forthcoming). Migrant Remittances and Information Flows:Evidence from a Field Experiment. World Bank Economic Review.

Bauer, T. and K. Zimmermann (1994). ‘Modelling International Migration: Economicand Econometric Issues’. In R. van der Erf and L. Heering (eds), Causes of InternationalMigration. Proceedings of aWorkshop, Eurostat, Luxembourg, 14–16 December. Brussels:Statistical Office of the European Communities.

CIEM (Central Institute of Economic Management) (2011). Characteristics of theVietnamese Rural Economy: Evidence from a 2010 Rural Household Survey in 12 Provincesof Viet Nam. Report. Ha Noi: Statistical Publishing House.

CIEM, DERG, and IPSARD (Central Institute of Economic Management, DevelopmentEconomics Research Group, Institute of Policy and Strategy for Agriculture and RuralDevelopment) (2013). Characteristics of the Vietnamese Rural Economy: Evidence from a2012 Rural Household Survey in 12 Provinces of Viet Nam. Report. Ha Noi: StatisticalPublishing House.

Elsner, B., G. Narciso, and J. J. J. Thijssen (2013). Migrant Networks and the Spread ofMisinformation. IZA Discussion Paper 7863. Bonn: Institute for the Study of Labour(IZA). Available at: <http://ftp.iza.org/dp7863.pdf>. Accessed 11 September 2015.

Gröger, A. andY. Zylberberg (2016). ‘Internal LaborMigration as a ShockCoping Strategy:Evidence from a Typhoon’. American Economic Journal: Applied Economics, 8(2): 123–53.

GSO (General Statistics Office of Vietnam) (2011). The 2009 Viet Nam Population andHousing Census: Migration and Urbanization in Viet Nam: Patterns, Trends and Differentials.Ha Noi: General Statistics Office of Vietnam.

Harris, J. R. and M. P. Todaro (1970). ‘Migration, Unemployment, and Development:A Two-Sector Analysis’. American Economic Review, 61: 126–41.

McKenzie, D., J. Gibson, and S. Stillman (2013). ‘A Land of Milk and Honey with StreetsPaved with Gold: Do Emigrants Have Over-Optimistic Expectations about IncomesAbroad?’ Journal of Development Economics, 102(C): 116–27.

Maimbo, S. M. and D. Ratha (2005). Remittances: Development Impact and Future Prospects.Washington, DC: World Bank.

Nguyen, D. L., K. Raabe, and U. Grote (2015). ‘Rural–Urban Migration, HouseholdVulnerability, and Welfare in Viet Nam’. World Development, 71: 79–93.

Nguyen, T. P., N. T. M. T. Tran, T. N. Nguyen, and R. Oostendorp (2008). Determinantsand Impacts of Migration in Viet Nam. Working Paper 01. Ha Noi: Development andPolicies Research Centre (DEPOCEN).

Phan, D. and I. Coxhead (2010). ‘Inter-Provincial Migration and Inequality duringViet Nam’s Transition’. Journal of Development Economics, 91(1): 100–12.

Stark, O. (1991). The Migration of Labour. Oxford: Basil Blackwell.United Nations Vietnam (2010). Internal Migration: Opportunities and Challenges forSocioeconomic Development in Viet Nam. Ha Noi: United Nations.

World Bank (2016). Vietnam 2035: Toward Prosperity, Creativity, Equity and Accountabil-ity. Washington DC: World Bank.


157

http://ftp.iza.org/dp7863.pdf

8

Information and CommunicationTechnology

Heidi Kaila

8.1 Introduction

Across the developing world during the 21st century an important character-istic of structural change has been the emergence of the information andcommunication technology (ICT) sector. This development has been particu-larly rapid in Viet Nam, where a true ICT revolution has taken place. Thischapter examines household ownership of ICT: the geographic and demo-graphic differences across ownership of different kind of ICT, especiallyphones and the Internet, and factors related to adoption.From a macroeconomic perspective, ICT relates to the literature on

endogenous growth, which studies how the development of technologythrough innovations determines growth in the long run.1 ICT is hence relatedto structural transformation through technological progress. There are at leasttwo ways through which ICT is related to structural transformation. First, theemergence of the ICT sector itself is part of a structural transformation pro-viding new opportunities for employment and entrepreneurship. Second, asICT can be viewed as a new general purpose technology, it relates to structuraltransformation indirectly via transforming existing sectors by improvingefficiency in communicating and acquiring information.The nature of improved information acquisition and communication

relates to themicroeconomic literature on the use of ICT to overcome informa-tionbarriers and to reduce information asymmetries. In the case ofmarkets, thismeans that ICT use can increase market efficiency.

1 For an overview, see Aghion and Howitt (1998).

Evidence on the effects of ICT on market efficiency using developing coun-try data is mainly focused around the benefits of mobile phones (Jensen 2007;Muto and Yamano 2009; Aker 2010; Fafchamps and Minten 2012; Aker andFafchamps 2015). The literature finds that the introduction of mobilephones has contributed to more efficient pricing (Jensen 2007) and reducedprice dispersion among both consumer and producer prices (Aker 2010;Aker and Fafchamps 2015), or increased market participation (Muto andYamano 2009).

In the case of Viet Nam, Nguyen and Schiffbauer (2015) find that internetuse in firms is positively correlated with higher productivity growth, evenmore so in firms that are doing e-commerce. To our knowledge there are noprevious studies on the benefits of the Internet or phones on a household levelin Viet Nam, nor on the drivers of adoption choices.

In this chapter, we find that phone ownership is almost universal in 2014,but the Internet is still a technology used by a smaller minority which is muchwealthier and more educated than the average population. Even thoughinternet use is still much less common than phone ownership, we findthat same factors—education, income, and wealth—drive both phone andinternet adoption. Lastly, we also find that even though this rapid expansionof ICT has taken place, mobile phones and the Internet are still not consideredas the most important sources of information among the VARHS households.

Infrastructure might have played a role in the adoption choices of thesetechnologies during the time span studied, but in 2014, the coverage of 2Gand 3G was universal, and therefore infrastructure constraints cannot fullyexplain these differences (VNPT 2015). Policies that were implemented inorder to increase competition have been instrumental in influencing theadoption of mobile services (Hwang, Cho, and Long 2009). The internetprovider market has also been subject to similar policies. According to aWorld Bank report (Tuan 2011) eleven enterprises had been granted licencesto build broadband internet infrastructure in 2011. In practice, however, onlythree companies had exercised this right on a national scale. These are state-owned enterprise Viettel, the Viet Nam Post and Telecommunications Group(VNPT) and EVN Telecom.2

A growing part of technology adoption literature demonstrates how newtechnologies spread though social networks: knowledge of benefits and waysof using new technologies can spread through neighbours and friends(Munshi 2004; Bandiera and Rasul 2006; Conley and Udry 2010; Oster andThornton 2012; BenYishay and Mobarak 2014). This by construction is par-ticularly true for technologies that have network benefits, of which ICT is a

2 EVN Telecom has since merged with Viettel (Vietnam News 2011).

Information and Communication Technology

159

perfect example.3 We do not investigate network effects in this chapter, butthese findings of the literature are good to keep in mind when interpretingour results.This chapter proceeds as follows. Section 8.2 presents descriptive statistics

on the geographical patterns of ICT adoption. Section 8.3 presents descriptivedemographic statistics on ICT ownership comparing adoption rates tonational statistics, and Section 8.4 studies the determinants of ownership.Section 8.5 discusses the role of ICT as a source of information, and finallySection 8.6 concludes.

8.2 Geographical Differences

In this section, we investigate geographical differences related to adopting newtechnologies.4 We have grouped the provinces into five categories according totheir region. The categories constitute the following provinces: Red River Delta(Ha Tay), North (Lao Cai, Phu Tho, Lai Chau, and Dien Bien), Central Coast(Nghe An, Quang Nam, and Khanh Hoa), Central Highlands (Dak Lak, DakNong, and Lam Dong), and Mekong River Delta (MRD) (Long An).Figure 8.1 panel A shows the average share of households per region that have

either a fixed-line phone or a mobile phone.5 There has been a tremendousincrease in phone ownership across the country, and also convergence acrossprovinces. In 2006, the share of households with phones ranged from 13 percent on average in the Northern provinces to 28 per cent at the MRD (Long Anprovince). The gap has narrowed ranging from 87 per cent in the Central Coastprovinces to 95 per cent in the provinces in the Central Highlands in 2014.The development of television ownership is presented in Figure 8.1 panel

B. It follows a similar converging pattern to that of phones, with the differencethat the initial levels of ownership rates were much higher in 2006. As withphones, television ownership in the Northern provinces has caught up withthe rest of the country. Television and government-owned channels have adominant role in the Vietnamese media (BBG 2013). Television is consideredto be one of the most important information sources also among VARHShouseholds.6

Computer and internet adoption is presented in Figure 8.1, panels C and D,respectively. Surprisingly, there is no sign of geographical convergence, but a

3 See Bjorkegren (2015) for a welfare analysis of the adoption of mobile phones in Rwanda.4 See Chapter 1 for an international comparison of broadband subscription rates.5 In VARHS the question about phone ownership refers to both fixed line and mobile phones,

hence we cannot differentiate between these two types. This is why we speak of ‘phones’throughout the chapter.

6 See Section 8.5 for details.


160

00.10.20.30.40.50.60.70.80.9

1

6 8 10 12 14

Shar

e o

f H

H

Year

Red River Delta North Central Coast

Central Highlands Mekong River Delta

00.020.040.060.080.1

0.120.140.160.180.2

6 8 10 12 14

Shar

e of

HH

Year



0.4

0.5

0.6

0.7

0.8

0.9

1

6 8 10 12 14

Shar

e o

f H

H

Year





00.05

0.10.15

0.20.25

0.30.35

0.40.45

0.5

6 8 10 12 14

Shar

e o

f H

H

Year

Panel A: Share of households with at least one phone

Panel C: Share of households with at least one computer Panel D: Share of households with access to internet

Panel B: Share of households with at least one color TV

Figure 8.1 Geographical distributions of technology ownershipSource: Author’s calculations based on VARHS dataset for 2006–14.

diverging pattern. In 2006, there were almost no computers in any of theprovinces and owning computers was still quite uncommon in 2014. Therehave been slight increases in computer ownership in all of the provinces, withthe most rapid increases taking place in the provinces in the Central High-lands and Red River Delta. In the Northern provinces, the increase has beenmuch slower on average. In all of the areas the share of households with atleast one computer is less than 20 per cent, so the development has overallbeen very moderate compared to that of phones.What we observe with computer ownership can in fact be considered as a

lower bound of computer use. Internet cafés, work, and education providehouseholds with opportunities for computer use, when they do not own acomputer themselves. Therefore, studying internet access, a variable that alsotakes into account use from work and from an internet café, might give us amore realistic view of computer use.Development in internet access (limited to categories ‘home’, ‘work’, or

‘internet café’) is presented in Figure 8.1 panel D. The sharpest increase hasbeen in the Northern provinces, where there was nearly zero access in 2006.Internet cafés are losing popularity, a factor that fully explains the decreasebetween 2012 and 2014 in the Central Highlands, Central Coast, and MRDprovinces.7 Since we do not observe all ways in which the Internet is used, forinstance use from a mobile phone, it is reasonable to assume that our internetmeasure is biased downwards. This is supported by the comparison betweenthis household-level measure and the commune-level internet access variablepresented in Chapter 3, Figure 3.9. We can see that commune-level internetavailability is increasing in all regions over the entire time period studied.

8.3 Descriptive Statistics of ICT Ownership

There are several factors that determine the choice of adopting new ICT. In theearly stages, infrastructure plays a key role: for instance, electricity or a tele-phone grid are necessary conditions for the adoption of a television and alandline telephone. In this section we study how households with and with-out a phone or an internet connection differ in terms of access to infrastruc-ture, wealth, income composition, and other characteristics.

7 The decrease in popularity of internet cafés also explains the slight decrease in internet use inthe Central Coast between 2006 and 2010. This is fully attributable to a decline in internet café usein the Quang Nam province. It is noteworthy to point out that the initial level of internet use in2006 was highest in this region, thus the decrease might reflect that they were ahead of others insubstituting internet cafés with other means of using the Internet, which are not captured in ourdata. Access from work and home are very low values when collapsed to a regional or provincelevel, but these categories are increasing over time in all regions.


162

8.3.1 Descriptive Statistics on Phone Ownership

From all the technologies studied in our dataset, the expansion of phones hasbeen the most striking one. Figure 8.2 illustrates a tremendously rapid expan-sion of phone ownership over 2006–14. Between 2006 and 2010 the averagenumber of phones owned by household was close to zero—in 2014 it was two.Between 2006 and 2008, the share of households that had at least one phone

0102030405060708090

0 1 2 3 4 6 7

Perc

enta

ge

Number of phones

2006

0

10

20

30

40

50

60

0 1 2 3 4 55 6 7

Perc

enta

ge

Number of phones

2008

0

5

10

15

20

25

30

35

40

0 1 2 3 4 5 6 7

Perc

enta

ge

Number of phones

2010

0

5

10

15

20

25

30

0 1 2 3 4 5 6 7

Perc

enta

ge

Number of phones

2012

0

5

10

15

20

25

30

35

0 1 2 3 4 5 6 7

Perc

enta

ge

Number of phones

2014

Figure 8.2 Number of phones owned by householdSource: Author’s calculations based on VARHS dataset for 2006–14.


163

doubled from 18.6 per cent to 38 per cent. In 2014, there was almost fullcoverage of phones, the share of household with at least one phone being 91per cent.In VARHS, we cannot differentiate between a fixed-line phone and a mobile

phone. However, it is reasonable to assume that the increases in phone owner-ship in VARHS are attributable to the increase in mobile phones, not fixed-lineconnections. According to the General Statistics Office (GSO) of Viet Nam (GSO2015a), on a national level the number of fixed-line connections has increasedby 9 per cent between 2006 and 2012. Over this same time period, the share offixed-line subscriptions of all subscriptions has decreased from 30.7 per cent to6.7 per cent, which confirms that most phones must indeed be mobile phones.The average number of phones owned by a household in VARHS has

increased almost sevenfold between the years 2006 and 2012 (from 0.25 to1.61 phones per household) and almost eightfold between the years 2006 and2014 (to 1.94 phones per household). A comparison to the national statisticsreveals that the adoption of mobile phones has been more rapid in VARHSprovinces than in Viet Nam in aggregate: the total number of all telephonesubscriptions has increased slower than in VARHS, nearly fivefold between theyears 2006 and 2014, from 28.5 to 141.2 million. During the same period, thenumber of mobile phone subscriptions has gone up more than sixfold (from19.7 to 131.7 million).Even though adoption rates have been higher, the VARHS provinces have

not yet caught up with the national average in 2014 due to the initially lowlevels of phone ownership in 2014. In 2006, nationally there were about 0.34subscriptions per capita compared to 0.25 phones per household in 2006 inVARHS provinces. If we compare these same numbers for the most recentperiod available, 2012, we can see that the number of phone subscriptions was1.59 per capita on a national level and 1.61 per household in the VARHSprovinces. Since the average size of a household in VARHS is 4.4 members, wecan clearly conclude that the VARHS provinces still have not caught upmobile phone penetration rates to the national average.8

Another aspect that we do not observe in the data is the quality of themobile phones used, whether they are newer smartphones or traditionalmobile phones. Anecdotal evidence suggests that quality smartphones arepopular and carry a special luxury status in Viet Nam—Apple experienced itslargest increase in iPhone sales in the world in Viet Nam in the first half of2014 (Heinrich 2014). Apple holds a dominant position in the country, unlikein China where the local cheaper brands dominate the market.

8 The national phone subscriptions per capita are author’s calculations based on GSO (2015a,2015b).


164

Hwang, Cho, and Long (2009) investigate the determinants of mobilephone services diffusion in Viet Nam during 1995–2006, until the beginningof our dataset. Their conclusion is that policies taken to open the market forcompetition has been a significant factor determining the diffusion of mobilephone services, due to new service providers entering the market and thesubsequent decrease in prices. As 3G and 2G have nationwide coverage inViet Nam (VNPT 2015), the infrastructure constraints should no longer play alarge role in the purchasing decision of a phone in 2014. The quality of thesignal might, of course, be weaker in more remote areas.

Table 8.1 describes the differences between households that still do nothave a phone in 2014 with those that have one. In 2014 only 9 per cent ofhouseholds did not yet have a phone.

Table 8.1 Mean comparison across households with and without a phone, 2014

No Phone Phone Difference

Household size 2.70 4.28 �1.58***Female hh head 0.47 0.22 0.26***Education per capita 4.45 8.56 �4.12***Number of children <15 0.44 0.77 �0.33***Total area owned 5243.38 7475.80 �2232.42**Total area owned per capita 1698.87 1855.74 �156.88Monthly hh income per capita 1316.54 2167.49 �850.95***Crop production last year per capita 3525.82 8884.96 �5359.15***Classified as poor 0.39 0.10 0.29***Income share wage 0.20 0.36 �0.16***Income share non-farm enterprises 0.04 0.13 �0.09***Income share crops 0.17 0.21 �0.04**Income share private transfers 0.26 0.09 0.17***Income share public transfers 0.20 0.08 0.12***Electricity 0.94 0.99 �0.05***Toilet 0.78 0.91 �0.13***Good water 0.81 0.86 �0.05*household has access to the Internet 0.06 0.31 �0.25***Number of motorcycles 0.42 1.47 �1.06***Number of motorcycles per capita 0.13 0.36 �0.23***Number of colour TVs 0.81 1.08 �0.27***Number of colour TVs per capita 0.44 0.31 0.13***Number of computers 0.02 0.16 �0.14***Number of computers per capita 0.00 0.04 �0.03***Distance all-weather road 1.90 1.82 0.08Distance People’s Committee 2.34 1.97 0.37*Distance public health care 2.30 1.95 0.35*Distance private health care 7.38 5.36 2.02*Distance primary school 1.90 1.72 0.19Distance crop buyer 1.25 1.15 0.10Trust (positive) 0.90 0.90 �0.00Trust (negative) 0.56 0.47 0.09**

Notes: Number of households with a phone 1,959, and without a phone 203. Significance levels: *p < 0.10,**p < 0.05, ***p < 0.01.

Source: Author’s computation based on VARHS dataset for 2014.


165

We can see that households vary tremendously over phone ownership.Households without a phone are small (2.70 members) compared to phone-owner households (4.28 members) and also to the average household (4.13members). The differences are likewise very large with respect to gender of thehousehold head: Almost half of the households without a phone are female-headed, compared to 22 per cent of households with a phone.Differences with respect to education level and income are also substantial.

Education in years per capita is almost twice as high for households withphones. Income per capita in households without a phone is 61 per cent ofthe income per capita in households with a phone. Furthermore, almost40 per cent of the households without a phone are classified as poor by theauthorities versus 10 per cent of households with a phone.The sources of income differ significantly as well: households without

phones rely heavily on transfers, both public and private. Transfers accountfor nearly half of the incomes of these households, when the respectivenumber for phone owners is 17 per cent. Of the actual income-generatingactivities, wage-earning activities are the most important income source forhouseholds without phones, followed by income from crop sales. Incomeshare of own non-farm enterprises is 4 per cent compared to 13 per cent inhouseholds with phones.In terms of remoteness and household infrastructure, households without a

phone are less likely to have electricity, but the difference is rather small,electricity being almost universal in 2014. Households with no phone arealso less likely to have a toilet,9 or access to good water.10 Finally, householdswithout phone are somewhat more remotely situated, but the differences arequite small.11

As 3G is universal, infrastructure constraints should not play a large role inthe purchasing decision of a phone. Very remote areas might still have aweaker signal. On the other hand, if a phone is mostly used to keep in touchwith other family members, it is possible that smaller households simply haveless demand for phones. However, given that these households are also sig-nificantly poorer, the decision not to buy a phone might be due to financialconstraints.We have also investigated phone ownership and ownership of other dur-

ables or technology to study the assumption that being ‘tech-savvy’ plays arole in the purchasing decision, keeping in mind the fact that wealthier

9 Variable toilet gets the value one if the household has a toilet, otherwise this is zero.10 Goodwater is a dummyvariable taking value one, if thewater comes from the following sources:

tap water (private or public), tank, bought water, water from deep drilled wells, or hand-dug andreinforced wells. Any other kind of source of water gets value zero.

11 Distance to public services, such as health-care facilities, all-weather roads, schools, andPeople’s Committees are used as measures for remoteness.


166

households can better afford any kind of durables and technology. House-holds without a phone own one-third of the number of motorcycles per capitarelative to households with a phone. In addition, they are less likely to haveinternet access,12 or to own a computer. Contrary to these findings, house-holds without phones own more televisions per capita than householdswith phones.

Given that households without phones are slightly more remotely located,purchasing a phone might actually be beneficial for these households, if aphone is used to gain information about markets or public services. Coupledwith the fact that households which do not own a phone very rarely haveaccess to the Internet or own a personal computer, they seem to have fewermeans to benefit from possibilities brought by ICT.

Since a phone functions as a means to keep in contact with one’scommunity, we have also investigated whether there are differences withrespect to social capital. It is plausible that households that have a centralrole in a social network, or stronger ties to their community, are more likely toown a phone. Our data has two variables related to trust, which enable us tostudy whether households not owning a phone do have less trust in their owncommunity, and therefore might be less inclined to purchase such technologythat allows them to keep contact with their community.

The variable ‘trust (positive)’ is a dummy that takes value one if the respond-ent agrees with the statement ‘most people are basically honest and can betrusted’. We observe no difference with respect to this general perception oftrust, 90 per cent of respondents in both groups agree with this statement.However, we do observe a significant difference in the variable ‘trust (nega-tive)’, which takes value one if the respondent agrees with the statement‘in this commune one has to be careful, there are people you cannot trust’. Inconclusion, households without a phone do tend to have slightly less trustin their community than phone owners: over one-half of the respondents inthis group agree with this statement.

8.3.2 Descriptive Statistics on Computer and the Internet

The ownership of personal computers for the years 2006–14 has increasedfivefold from 2.4 per cent in 2006 to 12.9 per cent in 2014, amoderate increasecompared to phone ownership.13 As smartphones have recently gained popu-larity in Viet Nam, it is possible that they complement the functionalities ofpersonal computers subsequently decreasing the demand for computers.

12 Internet use is a dummy getting value one if using Internet from home, work, or internet café.13 For details, see Kaila (2015), a related working paper.


167

According to Cimigo (2011), in 2011 the most important activity on theInternet was information gathering, which refers to the use of search sites andreading online news. The Internet was only secondarily used for communica-tion and entertainment.14 Therefore, it is reasonable to assume that oldmedia,which is under heavy government control (see BBG 2013), are challenged bynew information sources.Figure 8.3 illustrates the development of internet use among VARHS house-

holds comparing 2006 and 2014. There has been a large increase in the shareof households having internet access, but this development is less strikingthan that of mobile phones. Internet access is measured by the question ‘Doesanyone in your household have access to internet services? If so, wheremainly?’, with the response categories given in the figure. We can see that,altogether, 16.1 per cent of households had internet access (fromwork, home,or internet café combined) in 2006, and 28.4 per cent of households in 2014.The increase is fully attributable to the increase in access from home andworkplace that has gone up from close to zero to around 9 per cent. Simul-taneously, we observe a decrease in the category ‘access from internet café’from 13.9 per cent in 2006 to 10.8 per cent in 2014. The overall level of accesshas stayed the same since 2012, after which the decrease in access frominternet café has been compensated by access from home and from work onaggregate.

0 10 20 30 40Percentage

50 60 70 80

No access

Access from home

Access from workplace

Access from internet café

I don't know what internet is

2006 2014

Figure 8.3 Sources of internet access in 2006 and 2014Source: Author’s calculations based on VARHS dataset for years 2006 and 2014.

14 It is also noteworthy to point out that online banking has not been a phenomenon in VietNam as in eastern Africa. E-commerce is, however, starting to be an important part of internet use(see Cimigo (2011) for details about e-commerce and online banking in Viet Nam, and Mbiti andWeil (2016) on mobile banking in Kenya).


168

Compared to national development in internet subscription, the VARHSprovinces are really lagging behind. According to GSO, the national subscrip-tion rate for asymmetric digital subscriber line (ADSL) connections hasincreased more than eightfold during the period 2006 to 2012. At the sametime, and even going further to 2014, internet access in the VARHS provinces,as measured in our data, has increased by only 57 per cent. The difference isstriking, even though we do have several reasons to believe that these num-bers are not fully comparable.

First of all, national ADSL subscriptions take into account all the subscrip-tions by firms. Clearly, the demand has been higher in urban areas due to adifferent economic structure in large cities. It is only natural that in areaslargely dependent on agriculture, there is less demand for ADSL. Even thoughwe allow the households to report that they have access from work, we areunable to capture all internet use from work, since respondents could havechosen another category for internet access.

Second, there might be measurement error in our internet access measure.Internet use from a mobile phone might also have replaced internet cafésto some extent. As we do not observe any kind of access via mobile phone,it is reasonable to assume that our measure of internet access is biaseddownwards.

Even considering all these caveats, the difference between the eightfoldincrease in national ADSL subscriptions, compared to the 57 per cent increasein internet access, together with the low level of computer ownership, doesraise the concern that VARHS provinces are lagging behind the nationalaverage on internet access.

Even though internet use has increased relatively moderately, we canobserve from the data that knowledge about what the Internet is has increasedover 2006–14. In 2006, more than 70 per cent of respondents chose thecategory ‘I don’t know what Internet is’ and only 11.5 per cent reported thatthey ‘don’t have access’. In 2014, the share of respondents that chose theformer was just 16.8 per cent and of the latter 54.8 per cent. Hence, knowledgeabout what the Internet is has indeed increased, even though the householdshave not reported having internet access.

Next, in Table 8.2 we investigate differences across households that haveinternet access to those that do not in 2014, with respect to the same charac-teristics as in Table 8.1 on phones. The share of households with internetaccess was 28 per cent, that is 615 households, and the number of householdswithout access was 1,547.

It is worth pointing out that households with no phones represent a smallminority in 2014 (9 per cent), households without internet access are still alarge majority (72 per cent). However, the differences between users and non-users of these two technologies are qualitatively similar.


169

Households without internet access are also smaller and more often female-headed than households with the Internet, but these differences are very smallcompared to the difference in phone ownership. The gap in education percapita is also strikingly large; households without access have close to threeyears less education. In addition, income differences are very large: house-holds with the Internet earn 60 per cent more than households without theInternet. Again, we also observe that the value of crop production in per capitaterms is larger for households with internet access.Income share of wages is 46 per cent in households with internet access,

compared to 30 per cent in households with no internet access. Thisdifference is explained mostly by there being more households with noincome at all from wage labour among non-user households than among

Table 8.2 Mean comparison across households with and without internet access, 2014

No Internet Internet Difference

Household size 3.95 4.58 �0.62***Female hh head 0.26 0.20 0.06***Education per capita 7.40 10.13 �2.73***Number of children <15 0.76 0.66 0.10**Total area owned 7186.91 7465.62 �278.72Total area owned per capita 1898.56 1696.25 202.32Monthly hh income per capita 1787.87 2841.54 �1053.67***Crop production last year per capita 7775.76 9906.14 �2130.38*Classified as poor 0.16 0.05 0.11***Income share wage 0.30 0.46 �0.15***Income share non-farm enterprises 0.10 0.17 �0.06***Income share crops 0.22 0.17 0.05***Income share private transfers 0.12 0.06 0.06***Income share public transfers 0.10 0.06 0.04***Electricity 0.99 0.99 �0.01Toilet 0.88 0.96 �0.09***Good water 0.84 0.90 �0.06***Number of phones 1.74 2.81 �1.06***Number of phones per capita 0.48 0.64 �0.16***Number of motorcycles 1.16 1.93 �0.77***Number of motorcycles per capita 0.30 0.44 �0.14***Number of colour TVs 1.01 1.18 �0.17***Number of colour TVs per capita 0.33 0.29 0.05***Number of computers 0.03 0.45 �0.42***Number of computers per capita 0.01 0.11 �0.10***Distance all-weather road 2.04 1.29 0.75***Distance People’s Committee 2.10 1.78 0.32***Distance public health care 2.10 1.69 0.41***Distance private health care 6.37 3.50 2.86***Distance primary school 1.80 1.58 0.21**Distance crop buyer 1.27 0.86 0.41**Trust (positive) 0.89 0.92 �0.02Trust (negative) 0.48 0.48 �0.00

Notes: Number of households with internet access 615, without 1,547. Significance levels: *p < 0.10, **p < 0.05,***p < 0.01.

Source: Author’s computation based on VARHS dataset for 2014.


170

user households.15 Households without internet access are alsomore dependenton transfers and rely slightly more on agriculture as an income source.

With respect to remoteness, the differences between households with andwithout internet access are larger than between households with and withouta phone. Households without internet access are more remotely located. Withrespect to infrastructure, we observe very similar results to those of phoneownership: households without access to the Internet are less likely to have atoilet or access to good water.

An explanation might be simple. In 2014, not having a phone seems to beassociated with poverty, whereas not having internet access is certainly relatedto lower income, but also generally to relying more on agriculture as a sourceof income. Of the households without internet access, only 16 per cent have apoverty status, compared to 5 per cent of households with internet access.This difference is significant, but not nearly as large as the difference inpoverty status in phone ownership.

With respect to ownership of other technology, we find that householdswithout internet access have more than one phone less on average thanhouseholds with access, and the difference is significant even in per capitaterms. In addition, motorcycle ownership is significantly smaller, and unsur-prisingly, also the number of computers. Households with internet accesshave on average 0.45 computers at home compared to nearly zero amonghouseholds without access. Finally, with respect to trust measures, we cannotsay that households with and without internet access differ at all.

Internet access appears to be related to a wealthy lifestyle: working outsideagriculture, being more centrally located, and having higher income and edu-cation, factors that surely correlated with each other. These findings are also inline with internet use reports that also look at urban population (Cimigo 2011;BBG 2013).

8.4 Determinants of ICT Adoption

Descriptive results give us a good picture of the differences between house-holds that have access to ICT and households that do not. In this section wego a step further and examine which factors are correlated with phone andinternet adoption, when other factors are controlled for, and on the otherhand, which are not.

15 The statistically significant differences in several of the income share variables are largely dueto the difference in the share of households that have zero income in that category. For instance,with respect to wage income, the distributions are almost uniform in the positive values for bothhouseholds with and without access to Internet, but the share of households with zero wageincome is much larger among households with no internet access.


171

We regress phone and internet variables on a set of explanatory variables inthe following way:

techit ¼ αþ β1Xdemoit þ β2Incomeit þ β3Xinfra

it þ β4Xassetit þ β5Xdistance

it þ αi þ τt þ εit

Where techit is the number of phones per household in the statistical model ofSection 8.4.1, and the dummy for internet access in Section 8.4.2. Theexplanatory variables are grouped in the following way for the sake of illus-tration: Xdemo

it is a vector of demographic variables including household size,education per capita (in years), dummy taking value 1 if head of the householdis female, number of children less than 15 years of age and log of total landarea owned by household. Incomeit is the log income of the household, Xinfra

it

is a vector of infrastructure related variables: whether the household has elec-tricity, toilet or good water. Xasset

it is a vector of technological assets owned bythe household: number of motorcycles, colour TVs, and computers. InSection 8.4.1 it also includes a dummy for internet access, in Section 8.4.2 itincludes the number of phones. Xdistance

it is a vector of distance measures, αi arehousehold fixed effects (in Section 8.4.2 random effects are used instead), τt areyear dummies, εit is an error term and α is a constant. In some specifications wehave commune fixed effects instead of household fixed effects or randomeffects. All standard errors are clustered at the commune level.It is noteworthy to point out that the purpose of this exercise is not to make

causal claims but to investigate partial correlations holding constant otherfactors. Our regressions estimates cannot be interpreted as causal at least fortwo reasons: first, there might be unobservable variables that we have notbeen able to control for, which can confound the results. We have tacked thisproblem as well as possible by including household fixed effects or randomeffects as well as regional dummies (the details of the different model specifi-cations are explained alongside the results). Second, it is also plausible thattechnology ownership has an effect on our explanatory variables, in whichcase causality could go both ways. Finally, the linear regression assumes alinear functional form for the sake of simplicity, it is not founded in economictheory.16 Keeping in mind these restrictions to the interpretation of ourcoefficient estimates, the partial correlations reveal interesting relationshipsrelated to technology ownership and household characteristics nevertheless.

8.4.1 Determinants of Phone Ownership

In Table 8.3 we have investigated the determinants of phone ownership over2008–14, where the dependent variable is the number of phones in a household.

16 In the regressions related to internet access dummy, a probit model is also estimated.


172

Table 8.3 Dependent variable: number of phones

(1) (2) (3) (4) (5)

OLS OLS OLS OLS HH FE

Household size 0.144*** 0.096*** 0.097*** 0.095*** 0.126***(10.10) (9.07) (9.22) (8.46) (8.41)

Education per capita 0.090*** 0.054*** 0.053*** 0.051*** 0.032***(18.21) (12.65) (12.25) (11.11) (4.37)

Female hh head �0.067** �0.051* �0.053* �0.046 0.046(�1.97) (�1.76) (�1.80) (�1.45) (0.81)

Number of children <15 �0.142*** �0.106*** �0.105*** �0.108*** �0.130***(�8.24) (�7.09) (�7.06) (�6.84) (�5.45)

Total area owned (ln) 0.020** 0.011 0.013 0.001 0.008(2.09) (1.29) (1.41) (0.11) (0.56)

Hh income (ln) 0.388*** 0.213*** 0.211*** 0.199*** 0.152***(19.47) (12.06) (11.94) (12.39) (7.64)

Electricity 0.123* 0.011 �0.010 0.025 0.090(1.88) (0.17) (�0.16) (0.42) (1.24)

Toilet 0.247*** 0.170*** 0.163*** 0.127*** 0.073*(5.54) (4.42) (4.30) (3.20) (1.66)

Good water 0.108*** 0.043 0.038 0.031 0.048*(2.95) (1.40) (1.25) (1.23) (1.85)

household has access to theInternet

0.285*** 0.283*** 0.290*** 0.262***

(9.49) (9.45) (9.69) (8.26)Number of motorcycles 0.353*** 0.354*** 0.340*** 0.291***

(12.19) (12.18) (10.90) (8.60)Number of colour TVs 0.176*** 0.170*** 0.156*** 0.078**

(5.06) (4.95) (4.95) (2.52)Number of computers 0.393*** 0.394*** 0.363*** 0.304***

(8.78) (8.81) (7.62) (5.58)Distance all-weather road �0.003 �0.003 �0.003

(�1.39) (�1.35) (�1.22)Distance People’sCommittee

�0.002 �0.003 �0.005

(�0.55) (�1.09) (�1.14)Distance public health care 0.001 �0.004 0.000

(0.21) (�0.89) (0.06)Distance private health care �0.000 �0.000 0.000

(�0.70) (�0.02) (0.47)Distance primary school �0.005 �0.003 �0.002

(�1.05) (�0.66) (�0.45)Constant �4.050*** �2.143*** �2.086*** �1.780*** �2.300***

(�16.30) (�10.61) (�10.35) (�9.45) (�9.25)Observations 11,976 11,976 11,976 11,976 11,976Commune No No No Yes NoYear Yes Yes Yes Yes Yes

Notes: Dependent variable is number of telephones owned by a household. In columns 1-4 pooled, OLS is used, incolumn 5 household fixed effects is used. Standard errors are clustered at the commune level. Standard errorsin parentheses. Significance levels: *p < 0.10, **p < 0.05, ***p < 0.01.

Source: Author’s computation based on VARHS dataset 2008–14 (balanced panel).


173

Columns 1–4 show ordinary least squares (OLS) results, controlling for year fixedeffects. Column4 also controls for commune level time-invariant characteristics.Column 5 displays the results with household fixed effects and therefore thecoefficient estimates can be interpreted as the relationship between the changesin the explanatory variables and the change in the number of phones, holdingunobservable time-invariant household characteristics fixed.The results show that when controlling for a number of household charac-

teristics, household income has a statistically significant relationship withthe number of phones or with the purchasing decision of phones.17 Oneimportant determinant in the number of phones is household size. Whencontrolling for household size however, having more children has a negativecoefficient estimate. Hence, the amount of adult members in a household ispositively associated with the number of phones.The remoteness of the household is not a statistically significant determin-

ant of phone adoption, even when we do not control for commune fixedeffects. In addition, land size does not seem to play a role. What is interestingis that the ownership of other technology—computers, the Internet, televi-sions, and motorcycles—has a large positive relationship with the number ofphones and on the decision to acquire more. These are all proxies for wealth,but the result could also be interpreted as suggestive evidence that being tech-savvy could play a role in the ownership of phones.

8.4.2 Determinants of Internet Adoption

Table 8.4 presents regression results for the determinants of internet access overthe period 2008–14. Columns 1–4 show the results from pooled OLS. In column4,we are controlling for commune fixed effects that absorb all commune specifictime-invariant characteristics. Column 5 is a random effects model and column6 a probit model with random effects (marginal effects reported), as our internetaccess measure is a dummy. In all specifications, we control for time dummiesand cluster the standard errors at the level of the commune.We can see that the determinants of internet adoption are somewhat similar

to those of phones. Owning another type of technology has a strong positiveassociation with having internet access; owning a computer is the mostimportant one. Larger households are also more likely to have access (espe-cially those with more adult members) and education also seems to be adriving factor. Even though households that do not have internet access aremore remotely located, when controlling for other household characteristics,the distance measures are no longer statistically significant. As the take-up of

17 We also ran the analysis splitting income into different income sources. The effects werequalitatively very similar (results not reported here).


174

Table 8.4 Dependent variable: internet access

(1) (2) (3) (4) (5) (6)

OLS OLS OLS OLS RE RE PROBIT

Household size 0.023*** 0.014*** 0.015*** 0.015*** 0.015*** 0.020***(0.004) (0.003) (0.003) (0.003) (0.003) (0.003)

Education per capita 0.027*** 0.018*** 0.018*** 0.017*** 0.018*** 0.020***(0.002) (0.001) (0.001) (0.002) (0.001) (0.001)

Female hh head 0.018 0.014 0.013 �0.003 0.013 0.008(0.011) (0.010) (0.010) (0.011) (0.009) (0.009)

Number of children <15 �0.037*** �0.029*** �0.029*** �0.028*** �0.029*** �0.038***(0.006) (0.005) (0.005) (0.005) (0.004) (0.004)

Total area owned (ln) �0.004 �0.004 �0.003 0.001 �0.003 �0.002(0.003) (0.003) (0.003) (0.002) (0.002) (0.002)

Hh income (ln) 0.074*** 0.033*** 0.032*** 0.029*** 0.032*** 0.031***(0.007) (0.006) (0.006) (0.006) (0.005) (0.005)

Electricity �0.008 �0.006 �0.014 �0.022* �0.016 0.021(0.010) (0.010) (0.010) (0.012) (0.015) (0.022)

Toilet 0.012 0.002 �0.001 0.002 0.000 0.010(0.009) (0.008) (0.008) (0.009) (0.009) (0.010)

Good water 0.028*** 0.020** 0.018** 0.016* 0.019** 0.022**(0.009) (0.009) (0.009) (0.009) (0.008) (0.009)

Number of phones 0.037*** 0.037*** 0.037*** 0.037*** 0.022***(0.004) (0.004) (0.004) (0.003) (0.003)

Number of motorcycles 0.017*** 0.017*** 0.014*** 0.016*** 0.010**(0.005) (0.005) (0.005) (0.004) (0.004)

Number of colour TVs �0.003 �0.005 �0.010 �0.003 0.009(0.007) (0.007) (0.007) (0.008) (0.008)

Number of computers 0.333*** 0.333*** 0.312*** 0.323*** 0.196***(0.014) (0.014) (0.013) (0.011) (0.010)

Distance all-weatherroad

�0.000 0.000 �0.000 �0.001**

(0.000) (0.000) (0.000) (0.001)Distance People’sCommittee

�0.000 0.000 0.000 0.000

(0.001) (0.001) (0.002) (0.002)Distance public healthcare

�0.002 �0.001 �0.002 �0.002

(0.001) (0.001) (0.002) (0.002)Distance private healthcare

�0.000 �0.000 �0.000 �0.000***

(0.000) (0.000) (0.000) (0.000)Distance primary school 0.001 0.001 0.001 0.002

(0.001) (0.001) (0.001) (0.001)Constant �0.848*** �0.409*** �0.387*** �0.372*** �0.387***

(0.070) (0.062) (0.061) (0.063) (0.052)Observations 11,976 11,976 11,976 11,976 11,976 11,976Commune FE No No No Yes No NoYear FE Yes Yes Yes Yes Yes Yes

Notes: Dependent variable is internet access, 1=yes, 0=no. In columns 1-4 pooled, OLS is used, column 5 uses arandom effects model. In column 6 a probit model with random effects is used, the coefficients reported are marginaleffects. Standard errors are clustered at the commune level. Standard errors in parentheses. Significance levels: *p < 0.10,**p < 0.05, ***p < 0.01.

Source: Author’s computation based on VARHS dataset 2008–14 (balanced panel).


175

the Internet has not been as rapid in the VARHS provinces as in the countryoverall, the question of how to make the Internet lucrative and accessible tothe rural population arises.As more and more citizens are gaining access to the Internet, they are also

using it increasingly as a source of information, in communication, andas a platform for working. The rural areas should not be left out of thisdevelopment—the Internet and ICT in general could also open up possibilitiesin the rural economy if the infrastructure and knowledge is there. Moreover,as internet cafés are losing their popularity and internet access is happeningmore and more through mobile phones and gadgets, easy ways of accessingthe Internet should be available for households that do not have the means topurchase these technologies or are not fully aware of their possible benefits.

8.5 Information Sources

Even though the adoption of ICT, phones in particular, has been rapid, wedo not yet know how the VARHS households perceive ICT. Figure 8.4

0

5

10

15

20

25

30

Perc

enta

ge

Sources of information

Relat

ives,

frien

ds an

d ne

ighbo

urs

Comm

unity

bull

letin

boar

d

Comm

unity

loud

spea

kers

Loca

l mar

ket

Radio

Telev

ision

Exte

nsion

agen

ts

Busin

ess o

r wor

k asso

ciate

s

Mob

ile p

hone

Inte

rnet

Other

Other

gro

ups o

r mas

s org

aniza

tions

agen

ts

Newsp

aper

Agriculture Credit Market Policy

Figure 8.4 Most important sources of information, 2014Source: Author’s compilation using VARHS dataset 2014.


176

illustrates the answers to the question: ‘Which sources of information areimportant for you? Regarding the following issues, list up to three for eachissue.’ Hence, each household has often named more than one informationsource. The full descriptions of the categories are: agricultural production andextension; sources of credit and insurance; market information—such as jobs,prices of goods, or crops; and government policy changes.

We can see that the Internet still plays a minor role as one of the mostimportant information sources, as do mobile phones. This is the case eventhough available evidence on the Internet suggests that it is used mainly toacquire information (Cimigo 2011; BBG 2013). However, both the Internetand mobile phones can operate as a means of communicating with one’ssocial network, which is listed as a separate category and not only for obtain-ing information from outside sources.

Indeed the social network—friends, relatives, and neighbours—is the mostimportant information source for agriculture, credit, and market information.In light of this finding, it is not surprising that technology adoption choicesare also strongly affected by social networks as suggested by recent literature(Munshi 2004; Bandiera and Rasul 2006; Conley and Udry 2010; Oster andThornton 2012; BenYishay and Mobarak 2014).

For government policy, television is the most important source and it is alsoused extensively to acquire other information. For information aboutmarkets,the local market is an important source. For all the categories, more traditionalchannels of information-spreading, community bulletin boards, communityloudspeakers, and other groups and mass organizations (MOs), are still veryrelevant.

8.6 Conclusions

The spread of ICT has been rapid in VARHS provinces over 2006–14. Theincrease in the ownership of phones has been extraordinary and probablyeven faster than in the country on average—the median number of phones ina household has increased from zero to two in eight years. Even though grossdomestic product (GDP) growth has also been high in Viet Nam, it is worthpointing out that not all technological development has been as fast as thetake-up of ICT services. For instance, mechanization of agriculture has beenmodest in VARHS provinces compared to the development of ICT.18

Despite the large relative increases in internet access and computer owner-ship, the VARHS provinces are lagging behind the national average in owner-ship levels. As phones become more sophisticated, phone ownership might

18 For details see Kaila (2015).


177

aid this issue in the future, provided that smartphones are also gaining popu-larity in the VARHS provinces.In 2014, households that did not own phones were much more likely to be

female-headed, poor, and less educated than households with phones. Internetaccess is associated with a relatively wealthier lifestyle than phone ownershipin 2014. The determinants of purchasing a new phone or obtaining internetaccess are nevertheless qualitatively similar: over 2008–14 we observe thateducation and income are important factors driving ICT adoption. In addition,households that already have technology are more likely to acquire more.Even though there has been a rapid increase in the number ofmobile phones

owned and also in internet use between2006 and2014, VARHShouseholds didnot report using mobile phones or the Internet as a main source for acquiringinformation about agriculture, markets, credit, and policy in 2014. More trad-itional channels are stillmore popular. However, themain use of the Internet isto acquire information and to get access to news. Therefore, as the popularity ofICT increases, old information sources will surely be challenged.As households that already have technology seem to be acquiring more of

it, barriers to adopting technology might be related to not knowing aboutthe benefits. When ICT becomes part of people’s everyday lives, it can beexploited in various realms of life in such ways that were unimaginablebefore. Therefore, having access to technology could serve as a means toempowerment. Households without access to any ICT services might be atrisk of exclusion from economic activity and have less means to makeeconomic choices. It is thus crucial that the rural population can afford tokeep up with a certain level of ICT.Investigating whether there are barriers to access to ICT and knowledge

about its use is a challenge for policy makers and further research on thetopic. In a growing economy, knowledge about ICT will surely bring aboutpossibilities for the young generations of today.

Acknowledgements

I would like to thank VARHS synthesis volume workshop participants for their valuablecomments.

References

Aghion, P. and P. Howitt (1998). Endogenous Growth Theory. Cambridge, MA: MIT Press.Aker, J. C. (2010). ‘Information from Markets Near and Far: Mobile Phones and Agri-cultural Markets in Niger’. American Economic Journal: Applied Economics, 2(3): 46–59.


178

Aker, J. C. and M. Fafchamps (2015). ‘Mobile Phone Coverage and Producer Markets:Evidence from West Africa’. World Bank Economic Review, 29(2): 262–92.

Bandiera, O. and I. Rasul (2006). ‘Social Networks and Technology Adoption inNorthernMozambique’. Economic Journal, 116(514): 869–902.

BBG (Broadcasting Board of Governors) (2013). ‘A Snapshot of Life and Media Use inVietnam: Findings from the World Poll’. Broadcasting Board of Governors technicalreport. Available at: <http://www.bbg.gov/blog/2013/11/26/bbg-research-series-vietnam-media-use/>. Accessed 3 May 2016.

BenYishay, A. and A. M. Mobarak (2014). Social Learning and Communication. NBERWorking Paper 20139. Cambridge, MA: National Bureau of Economic Research, Inc.

Bjorkegren, D. (2015). ‘The Adoption of Network Goods: Evidence from the Spread ofMobile Phones in Rwanda’. SSRN. Available at: <http://ssrn.com/abstract=2616524>.Accessed 3 May 2016.

Cimigo (2011). ‘VietnamNetCitizens Report 2011: Internet Usage and Development inVietnam’. Technical report, Cimigo.com. Available at: <http://www.cimigo.com/en/download/research_report/348>. Accessed 3 May 2016.

Conley, T. G. and C. R. Udry (2010). ‘Learning about a New Technology: Pineapple inGhana’. American Economic Review, 100(1): 35–69.

Fafchamps, M. and B. Minten (2012). ‘Impact of SMS-Based Agricultural Informationon Indian Farmers’. World Bank Economic Review, 26(3): 383–414.

GSO (General Statistics Office of Viet Nam) (2015a). Transport, Postal Services andTelecommunications Statistics. Available at: <http://www.gso.gov.vn/default_en.aspx?tabid=473&idmid=3&ItemID=15993>. Accessed 3 May 2016.

GSO (General Statistics Office of Viet Nam) (2015b). Population and EmploymentStatistics. Available at: <http://www.gso.gov.vn/default_en.aspx?tabid=467&idmid=3&ItemID=15748>. Accessed 3 May 2016.

Heinrich, E. (2014). ‘Apple’s Hottest Market? Vietnam’. Fortune. Available at: <http://fortune.com/2014/09/10/apple-hottest-market-world-vietnam/>. Accessed 3May 2016.

Hwang, J., Y. Cho, and N. V. Long (2009). ‘Investigation of Factors Affecting theDiffusion of Mobile Telephone Services: An Empirical Analysis for Vietnam’. Tele-communications Policy, 33(9): 534–43.

Jensen, R. (2007). ‘TheDigital Provide: Information (Technology), Market Performance,andWelfare in the South Indian Fisheries Sector’.Quarterly Journal of Economics, 122(3):879–924.

Kaila, H. (2015). ‘Comparing the Development of Agricultural Technology and Informa-tion Technology in Rural Vietnam’. WIDERWorking Paper 2015/091. Helsinki: UNU-WIDER.

Mbiti, I. and D. N. Weil (2016). ‘Mobile Banking: The Impact of M-Pesa in Kenya’. InS. Edwards, S. Johnson, and D. N. Weil (eds), African Successes: Modernization andDevelopment, vol. 3. Chicago: University of Chicago Press.

Munshi, K. (2004). ‘Social Learning in a Heterogeneous Population: Technology Diffu-sion in the IndianGreen Revolution’. Journal of Development Economics, 73(1): 185–213.

Muto, M. and T. Yamano (2009). ‘The Impact of Mobile Phone Coverage Expansion onMarket Participation: Panel Data Evidence from Uganda’.World Development, 37(12):1887–96.


179

http://www.bbg.gov/blog/2013/11/26/bbg-research-series-vietnam-media-use/

http://www.bbg.gov/blog/2013/11/26/bbg-research-series-vietnam-media-use/

http://ssrn.com/abstract=2616524

http://www.cimigo.com/en/download/research_report/348

http://www.cimigo.com/en/download/research_report/348





http://fortune.com/2014/09/10/apple-hottest-market-world-vietnam/

http://fortune.com/2014/09/10/apple-hottest-market-world-vietnam/

Nguyen, H. andM. Schiffbauer (2015). ‘Internet, Reorganization, and Firm Productivityin Vietnam’. Background paper for the World Development Report 2016, WorldBank, Washington, DC.

Oster, E. and R. Thornton (2012). ‘Determinants of Technology Adoption: PrivateValue and Peer Effects in Menstrual Cup Take-Up’. Journal of the European EconomicAssociation, 10(6): 1263–93.

Tuan, T. M. (2011). ‘Broadband in Vietnam: Forging Its Own Path’. Technical report,World Bank. Available at: <http://www.infodev.org/infodev-files/resource/InfodevDocuments_ 1127.pdf>. Accessed 3 May 2016.

Vietnam News (2011, December). ‘EVN Telecom, Viettel to Merge’. Available at:<http://vietnamnews.vn/Economy/218525/evn-telecom-viettel-to-merge.html>.Accessed 3 May 2016.

VNPT (Viet Nam Post and Telecommunication Group) (2015). About VNPT, NetworkCapacity. Available at: <http://www.vnpt.vn/Default.aspx?tabid=212&IntroId=268&temidclicked=268>. Accessed 3 May 2016.


180

http://www.infodev.org/infodev-files/resource/InfodevDocuments_1127.pdf

http://www.infodev.org/infodev-files/resource/InfodevDocuments_1127.pdf

http://vietnamnews.vn/Economy/218525/evn-telecom-viettel-to-merge.html

http://www.vnpt.vn/Default.aspx?tabid=212&e_x0026;IntroId=268&e_x0026;temidclicked=268

http://www.vnpt.vn/Default.aspx?tabid=212&e_x0026;IntroId=268&e_x0026;temidclicked=268

9

Social and Political Capital

Thomas Markussen

9.1 Introduction

This chapter investigates the evolution of different dimensions of social cap-ital in rural Viet Nam between 2006 and 2014 and models the relationshipbetween social capital and income at the household level. The literature onsocial capital distinguishes between three types of social ties: bonding (within-group ties), bridging (between-group ties), and linking (ties to people inpower). Linking social capital is sometimes referred to as ‘political capital’and I adopt this terminology here (Woolcock and Narayan 2000).

There is a complex, two-way relationship between economic developmentand these different dimensions of social capital. First, social capital affectsdevelopment. Some types of social capital facilitate economic growth andsophistication, while others are barriers to development. In particular, bridg-ing social capital facilitates interactions between strangers and therebyhelps to develop a sophisticated division of labour (e.g. Knack and Keefer1997). On the other hand, bonding and linking social capital may strengthenexclusivism and create biases in access to economic resources (only the‘insiders’ get a piece of the action), which in turn slows down economicgrowth. Second, economic development affects the structure of social capital.The growing need to interact with people from outside one’s own communityleads to a strengthening of bridging relative to bonding social capital. Formal-ized associations (political parties, trade unions, sports clubs, etc.) tend to partlyreplace informal associations (kinship ties, neighbourhood relations, etc.).

The ambiguous relationship between social capital and economic develop-ment is to someextent reflected inprevious studies of social capital inVietNam.WhileNewman, Tarp, and van den Broeck (2014) showpositive effects of socialcapital, measured as information sharing and membership of the Women’sUnion, on household savings (and thereby possibly on development),

Markussen and Tarp (2014) show that ‘linking’ social capital, in the form ofinformal ties between farmers and local government officials, distorts the dis-tribution of credit, monetary transfers, and agricultural investment. Similarly,Newman and Zhang (2015) report that politically connected households haveeasier access to public benefits than other households, and Kinghan andNewman (2015) find that politically connected families are more likely thanothers to establish a non-farm enterprise.This chapter uses the Viet Nam Access to Resources Household Survey

(VARHS) to investigate social and political capital in Viet Nam from differentangles. Due to changes in the VARHS questionnaire between 2006 and 2008,many of the analyses omit results for 2006. The chapter first presents descrip-tive statistics on the distribution of different dimensions of social capitalacross regions and socioeconomic groups and how this distribution has devel-oped over time. It then goes on to present regression analyses of the relation-ship between social capital and household income. These analyses showsignificant effects of several different aspects of social capital on householdincome. For example, Communist Party membership, trust in strangers, andinformal connections all affect income positively. On the other hand, I findno effect of membership in ‘mass organizations’ (MOs), such as the Women’sand Farmers’ Unions.

9.2 Communist Party Membership

Markussen and Tarp (2014) find that personal connections to local govern-ment officials, a form of political capital, strengthen land property rights andaccess to credit and transfers. This section focuses on another primary sourceof political capital in Viet Nam, namely membership of the Communist Party.In a one-party, highly activist state (‘totalitarian’ in many ways still seems anadequate description), the potential importance of Partymembership is obvious.Note thatmembership of the Party is far from open to all. Applicants go througha lengthy period of monitoring by current Party members and it is generally aprivilege reserved for a minority of the population (see Markussen et al. 2014for an analysis of the effects of Party membership on subjective wellbeing).Figure 9.1 (panel a) shows the share ofhouseholdswith at least onePartymemberacross five different regions and over time.1

The figure shows that the share of households with Party members increasedfrom a bit more than 7 per cent in 2008 to 11 per cent in 2014. This partlyreflects the fact that households are growing older and that this increases the

1 The VARHS questionnaire section on membership of the Party and other groups was changedbetween 2006 and 2008 and, for that reason, results for 2006 are omitted.


182

probability ofmembership. However, evenwhen the age of the household headis controlled, the difference between average membership in 2008 and 2014 isstill significant, indicating that the Party has somewhat expanded its member-ship base. Party membership is significantly more prevalent in the North thanin other regions. This is not surprising since the North is the traditional heart-land of the Party, but it is interesting to note that the strongest rate of growth

0

2

4

6

8

10

12

14

16

Panel a: By region


2008 2010Year

2012 2014

0

5

10

15

20

25

2008 2010Year

2012 2014

Perc

enta

ge

of

HH

wit

h a

tlea

sto

ne

mem

ber

Perc

enta

ge

of

HH

wit

h a

tlea

sto

ne

mem

ber


Richest quintile Second-richest Middle

Second-poorest Poorest quintile

Figure 9.1 Communist Party membershipNotes: N = 2,162 households (Each household is observed four times, so the total number ofobservations is 4 � 2,162 = 8,648). Income quintiles are calculated on the basis of per capitaincome and are defined ‘within year’, i.e. the sample is divided in five groups of equal size withineach year.Source: Author’s calculations based on VARHS 2008–14.


183

between 2008 and 2004 is observed in the Mekong River Delta (MRD), wherethe share of households with Party members more than doubles.Figure 9.1 (panel b) shows the distribution of Party membership across

income quintiles (VARHS collects highly detailed and comprehensive dataon household income). The figure reveals an extremely strong income gradi-ent in Party membership. Membership is four to seven timesmore common inthe richest quintile than in the poorest, with no convergence between 2008and 2014 (if anything, the 2014 results indicate the opposite). There may bedifferent reasons for this. Certain personal characteristics, such as educationand entrepreneurship, may affect both income and Party membership. Alter-natively, income may be a formal or informal criterion for Party membership,and/or Party membership is a cause of high income. The regression analysis inSection 9.7 throws more light on these issues. For now, it is sufficient to notethe tension between the egalitarianism of communist ideology and the socio-economic profile of Party members in rural Viet Nam.

9.3 Mass Organizations

Apart from the Party, the most important type of formal associations inrural Viet Nam are the so-called ‘MOs’, which include the Women’s Union,Farmers’Union, Youth Union, and Veterans’Union. Membership is voluntary,but MOs are closely linked with the state and sometimes participate in localgovernment decision-making. For example, in some communes the women’sand farmers’ unions participate in screening applicants for government-sponsored loans, for example, from the Bank for Social Policies (VBSP). If wedistinguish between ‘state’, ‘market’, and ‘civil society’ as the primary spheres ofsocial activities outside the family, social capital can be viewed as a measureof the strength of civil society. However, due to the strong links between MOsand the state, it is probably more relevant to view vibrant MOs as indicating astrong state, rather than a strong civil society. Nevertheless, group activitiesmay well be a source of bridging as well as linking social capital and theevolution and distribution ofMOmembership is therefore interesting to inves-tigate. Figure 9.2 (panel a) shows the average number of different MOs house-holds belong to, by region and over time.The figure shows that households are, on average, members of about 1.3

different MOs, with a slight increase from 2008 to 2014. About 75 per cent ofhouseholds aremembers of at least oneMO. TheMRD stands out as the regionwith by far the fewest MO memberships. Again, the most obvious interpret-ation is to view this as a legacy of the different histories of the communistmovement in the North and the South. However, it is interesting to note thatthere were important differences in the social structure of villages in the


184

northern and the southern deltas even before the advent of communism(in fact, even before colonialization). In particular, because of much lowerpopulation densities in the Mekong than in the Red River Delta, migrationwas more common in the South, which in turn meant that villages were lesstightly knit communities and that values were more individualistic (Gourou1936; Popkin 1979). It is possible that these historical differences are, to someextent, reflected in current social activities.

00.20.40.60.8

11.21.41.61.8

2

Panel a: By region


2008 2010Year

2012 2014

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

2008 2010 2012 2014

Mea

n n

umb

er o

f d

iffe

ren

t m

ass

org

.th

e H

H b

elo

ng

s to

Mea

n n

umb

er o

f d

iffe

ren

t m

ass

org

.th

e H

H b

elo

ng

s to


Year

Richest Second-richest Middle Second-poorest Poorest

Figure 9.2 Mass organization membershipNote: N = 2,162 households.Source: Author’s calculations based on VARHS 2008–14.


185

Figure 9.2 (panel b) shows MO membership by income quintile. In markedcontrast with the results on Party membership (Figure 9.1 (panel b)), there isno strong income gradient in membership of MOs. The Party is exclusive,MOs are inclusive.

9.4 Other Voluntary Associations

Now consider voluntary groups other than MOs. These include businessassociations, credit groups, religious groups, sports and cultural groups, groupsfor the elderly, and a number of other groups. Figure 9.3 compares frequencyof membership in, respectively, MOs and other voluntary groups.The figure shows that membership of MOs is more common than member-

ship of other groups by an order ofmagnitude, documenting thatMOs continueto dominate associational life in rural Viet Nam.However, the relative increase in non-MO membership from 2008 to 2014

(42 per cent) is much higher than the increase for MOs (11 per cent). Hence,some amount of convergence is perhaps underway. This is potentially veryinteresting, since the growth of non-MO voluntary groups could represent animportant step in the development of an independent civil society in VietNam. However, it is important to note that the growth in non-MO member-ship since 2008 is largely the result of growing membership in ‘groups for theelderly’. This growth is only partly explained by ageing of respondents. In alinear regression, which controls for age of the household head, membershipof non-MO groups is still significantly higher in 2014 than in 2008. Hence, the

0

0.2

0.4

0.6

0.8

1

1.2

1.4

2008 2010Year

2012 2014Mea

n n

umb

er o

f vo

lun

tary

gro

ups

per

HH

Mass organizations Non-mass organizations

Figure 9.3 Membership of mass organizations and other voluntary groupsNote: N = 2,162 households.Source: Author’s calculations based on VARHS 2008–14.


186

observed growth in non-MO membership is genuine. Still, it is unclearwhether these groups are able, for example, to play a role in holding govern-ment accountable, similar to the function of civic associations in northernItaly that Putnam (1993) famously described.

Figure 9.4 shows the development of non-MO memberships by region andincome quintile, respectively. It is notable that the average number of mem-berships has increased in all regions. Non-MOs are more common in northernthan in southern areas. Since these associations are not directly controlled

0

0.1

0.2

0.3

0.4

0.5

0.6

Panel a: By region


2008

Mea

n n

umb

er o

f n

on

-mas

s o

rg. V

olu

nta

ryg

roup

s th

e H

H b

elo

ng

s to

Mea

n n

umb

er o

f n

on

-mas

s o

rg. V

olu

nta

ryg

roup

s th

e H

H b

elo

ng

s to

2010Year

Year

2012 2014

0.5

00.05

0.10.15

0.20.25

0.30.35

0.40.45

2008 2010 2012 2014

Red River Delta North Central CoastCentral Highlands All Mekong River Delta


Figure 9.4 Non-mass organization membershipNote: N = 2,162 households.Source: Author’s calculations based on VARHS 2008–14.


187

by the communist movement, this can be said to go against the view thatassociational activities are driven only be the degree of communist dominance,which is surely stronger in the North than in the South. On the other hand, itis well in line with the view that northern villages are more ‘communitarian’than southern villages.Figure 9.4 (panel b) shows that, in 2008 and 2010, membership of non-MO

groups was more common in richer than in poorer households. However, thisdifference appears to have disappeared in 2012 and 2014, perhaps because theexpanding groups for the elderly cater to poor as well as to rich households.

9.5 Trust

It is unclear whether levels of voluntary group activity in Viet Nam are validmeasures of social capital, because of the strong links between the biggestgroups and the state. Therefore, attitudinal indicators, such as those measur-ing ‘trust’ are particularly interesting to investigate in a country such as VietNam. Measures of ‘generalized trust’, that is, trust in unspecific ‘strangers’,rather than specific groups or individuals, are commonly used as measures ofbridging social capital (e.g. Knack and Keefer 1997; Alesina and La Ferrara2002). The VARHS contains two such questions. The first asks respondentswhether they agree with the statement ‘most people are basically honest andcan be trusted’. The second asks about the statement ‘in this commune onehas to be careful, there are people you cannot trust’. Because the secondquestion refers to ‘this commune’, it is perhaps debatable whether it measuresbridging or bonding social capital (generalized or group-specific trust). However,since the number of inhabitants in a commune is about 5,000 on average,most residents in one’s commune are strangers in the sense that the respond-ent does not personally know them well. Therefore, I regard the question as ameasure of generalized trust and I combine answers to the two questions in anindex of trust. Figure 9.5 shows the share of respondents who agree with eachof the statements described earlier in this section.2

In general, the results show a very slight decrease in the share of respond-ents agreeing with the first statement (‘people are basically honest and can betrusted’) and a stronger decline, especially since 2008, in the share agreeingwith the second statement (‘one has to be careful . . . ’). Overall, this may betaken as evidence of a moderate increase in generalized trust. This may either

2 Note that, in 2006, two additional questions were inserted in the questionnaire between thefirst and the second of the questions discussed earlier in this section. These inserted questions wereremoved in 2008 and in later years. This may affect answers to the second question. In particular,the increase in the share who ‘agree’with the second statement from 2006 to 2008may reflect this.


188

be a cause or an effect of economic development, but in any case it should beviewed as good news. Generalized trust paves the way for economic special-ization and development.

In Figure 9.6, the two trust measures are collected in an index. Figure 9.6(panel a) shows the share of respondents who agree with the first statementand disagree with the second, by region and over time. Results again show anoverall increase in trust, especially since 2008. The difference between 2008and 2014 is highly statistically significant. The pattern across regions is rathermessy, with no clear trends emerging.

Figure 9.6 (panel b) shows the average score on the generalized trust indexby income quintile. There is no strong correlation between income and trust.It is curious that the order of the richest and poorest groups is completelyreversed between 2008 and 2014, but it is probably too early to draw strongconclusions from this result.

9.6 Family Ties

It is well documented that family ties are strong in Viet Nam. For example, the2001 World Values Survey (WVS) in Viet Nam asked respondents about theimportance of different ‘life domains’. A total of 82 per cent of respondents saythat the family is ‘very important’. Some 57 per cent regard ‘work’ as being inthe same category, while only 22 per cent rank ‘friends’ as very important(Dalton et al. 2002). Results from VARHS show that transactions with relatives

30

40

50

60

70

80

90

100

2006 2008 2010 2012Year

"Most people are basically honest and can be trusted""In this commune one has to be careful, there are people you cannot trust"

Perc

enta

ge

of

‘ag

ree’

2014

Figure 9.5 Generalized trust and mistrustNote: N = 2,162 households.Source: Author’s calculations based on VARHS 2006–14.


189

play a large role in, for example, land rental markets and in terms of gettingaccess to emergency funding (see later in this section). This suggests that stocksof ‘bonding social capital’ are high in rural Viet Nam. This is a strength when,for example, it comes to insuring households against negative shocks. However,as the economy develops, there is a growing need to interact with strangers andother non-kin. Therefore, we would expect a gradual decline over time in the

0

10

20

Perc

enta

ge

Perc

enta

ge

30

40

50

60

70

80

Panel a: By region


2006 2008 2010 2012 2014

30

35

40

45

50

55

60

2008 2010 2012 2014

Year

Year



Figure 9.6 Generalized trustNotes: N = 2,162 households. The figure shows the share of respondents who: (a) agree with thestatement ‘most people are basically honest and can be trusted’; and (b) disagree with the state-ment ‘in this commune one has to be careful, there are people you cannot trust’.Source: Author’s calculations based on VARHS 2006–14.


190

importance of family ties for economic transactions, as bonding social capital isreplaced or supplemented by a growing stock of bridging social capital. Thissection tests whether there is any support for this hypothesis in two types oftransactions: (a) emergency loans; and (b) land rentals.

VARHS asks respondents: ‘if you were in need of money in case of anemergency, who outside of your household could you turn to who would bewilling to provide this assistance?’ Slightly over 90 per cent list at least onesuch person. Respondents are asked to provide details about the three most

40

45

50

55

60

65

70

75

80

85

90

Panel a: By region


2008 2010Year

Year

2012 2014

50

55

60

65

70

75

80

2008 2010 2012 2014

Perc

enta

ge

of

hel

op

ers

wh

oar

e re

lati

ves

Perc

enta

ge

of

hel

op

ers

wh

oar

e re

lati

ves



Figure 9.7 Share of financial helpers who are relatives of the respondentNote: N = 3,849 financial helpers in 2008 (slightly more in in later years).Source: Author’s calculations based on VARHS 2008–14.


191

important helpers, for example, whether they are relatives or not. Figure 9.7shows the average share of financial helpers who are relatives of the house-hold, focusing on the three most important helpers (we only have detaileddata on the three most important helpers in each household). Again, changesin question formulation lead us to leave out results for 2006.Results show that the share of financial helpers who are relatives is about

70 per cent and, more importantly, that there is no decline in this share overtime. In fact the share of helpers who are relatives increases from 65 per centin 2008 to 75 per cent in 2014, a statistically significant difference, also whenage of household head is controlled for in a linear regression (not shown).Reliance on relatives for financial assistance is highest in the Red River Deltaand lowest in the Central Highlands (in three out of four years), possiblybecause many residents in the Central Highlands are migrants, who live faraway from their relatives.Figure 9.7 (panel b) shows the share of financial helpers who are relatives by

income quintile. Results show that there is virtually no correlation betweenincome and the importance of relatives for financial assistance. So, reliance onrelatives for informal insurance is not a peculiar characteristic of poor house-holds or backward regions, nor does this type of reliance show any signs ofdeclining over time.In Figure 9.8, I turn to the land rental market and consider the share of

tenants who are relatives of their landlord. This analysis is conducted atthe land plot level. About 8 per cent of the plots owned by households are

40

50

60

70

80

90

100

2006 2008 2010

Year

Perc

enta

g o

f te

nan

ts/b

orr

ow

ers

wh

oar

e re

lati

ves

of

the

lan

dlo

rd

2012 2014

Plots rented out Plots lent out for free

Figure 9.8 Share of rented plots where the tenant is a relative of the landlordNotes: This analysis is conducted at the plot level. N = 261 plots rented out, 294 plots lent out forfree in 2006; 539 plots rented out and 497 plots lent out for free in 2014 (intermediate numbers ofobservations in 2008–12).Source: Author’s calculations based on VARHS 2006–14.


192

rented out. For the plots rented out, the figure shows the share with a tenantwho is a relative of the landlord. The figure distinguishes between rentalagreements where a strictly positive rental fee was paid (in cash or kind),and arrangements where the land was lent out for free. Because the numberof plots rented out is relatively small, I do not break these results up by regionand income quintile. Results show that family ties are of primary importancein land rental markets. Unsurprisingly, this is especially true for plots lent outfor free. About 80 per cent of such agreements are between relatives. It is moreremarkable that even for genuine, rental agreements, where a rental fee ischarged, more than 50 per cent of contracts are between relatives. Even moreremarkably, there is no detectable decline in this share over time.

Hence, I find no support for the hypothesis of declining reliance on familyties in economic transactions. The importance of kinship relations in ruralViet Nam appears to be remarkably robust to economic development andwe may cautiously predict that the structure of economic transactions willcontinue to be shaped by family ties for a long time to come.

9.7 The Private Returns to Social Capital

One of the more straightforward and comprehensive ways to study the eco-nomic effects of social capital with household survey data is to model theeffects of the various dimensions of social capital on household income, as inthe much-cited paper by Narayan and Pritchett (1999) titled ‘Cents andSociability’. This section does that. Social capital may increase incomethrough several different channels. First, social capital helps groups solvecollective action problems, such as maintenance of irrigation systems, coord-ination of crop choice, joint marketing of agricultural output, and so on. Thisincreases income for all group members. At the individual level, networkspotentially help households in gaining access to good jobs or to cheapersupplies of credit and labour, thereby increasing their ability to invest andto profit from their businesses. Also, social capital is a source of insurance.Well-insured households are more willing to undertake risky investments,which may increase their income. Markussen and Tarp (2014) show thatpolitical capital increases the security of land property rights, which, in turn,is an important driver of agricultural investment and income.

Several caveats are in order. First, the model estimates the private returns tosocial capital. Private returns do not necessarily equal social returns. Forexample, a positive return to Communist Party membership does not implythat overall economic growth could be increased by expanding membership.More likely, such an effect picks up redistribution from non-members tomembers (although of course the Party may also be a forum that facilitates


193

solutions to collective action problems and thereby yield a positive, socialreturn). On the other hand, it is difficult to imagine negative externalities tohigher levels of generalized trust. Therefore, a positive, individual level effectof trust is more likely to reflect a positive, aggregate level effect also. Second,social capital may affect household welfare through other channels thanprivate income. First, strong social ties are a goal in themselves and not simplya means to material gain. Second, social capital may increase production ofcollective goods (e.g. crime prevention, public infrastructure), which is notincluded inmeasures of private income. Third, social capital may allow peopleto access consumption goods at lower prices than otherwise (as when neigh-bours share a harvest of fruit), leading to a direct effect of social capital onhousehold consumption.These things being said, total income is a relatively comprehensive measure

of the economic success of the household and it is interesting to see how thismeasure depends on the different aspects of social capital.I estimate models of the following type:

ln Yit ¼ S0itβþX

0itγþ αi þ φt þ εit

where Yit is real per capita income in household i in year t. S is a vector of socialand political capital measures. X is a set of control variables. αi is a householdfixed effect and φt is a year-dummy.εit is an error term, allowed to be correlatedwithin communes, the primary sampling unit of the VARHS. β and γ arevectors of parameters to be estimated. In the set of social capital measures,I include the variables discussed earlier: Communist Party membership,membership of MOs and other voluntary groups, the number of individualswilling to lend money in case of an emergency (‘financial helpers’), andscore on the trust index. Based on the findings in Markussen and Tarp(2014), Kinghan and Newman (2015), and Newman and Zhang (2015), ameasure of having a household member, relative or friend who is a localgovernment official is also included. As discussed, this is a measure oflinking social capital, or political capital.In the set of control variables, I distinguish between exogenous variables

(age, gender, schooling, and ethnicity of the household head) and potentiallyendogenous variables (number of working-age household members—thosebetween 15 and 65—and household assets). The set of asset variables includes,first, the amount of irrigated land. Irrigated rather than total land holdings areused because quality of land is often at least as important as quantity, andaccess to irrigation is a main determinant of land quality. Second, holdings ofa number of non-land assets are also included (numbers of, respectively, cows,buffaloes, telephones, bicycles, motorbikes, pesticide sprayers, and cars).Assets and household size are potentially endogenous in the sense that effectsof social capital on income may operate through these variables. For example,


194

social capital may ease access to credit, which in turn leads to faster assetaccumulation. Social capital may affect the number of working-age householdmembers by affecting the possibilities for and incentives to move out of orinto the household. Therefore, these variables are omitted from most of theregressions presented in this section. On the other hand, these factors mayalso be viewed as omitted third variables that affect both social capital andincome and for that reason they are included in some regressions.

One of the main difficulties of estimating the returns to social capital is thathouseholds with high and low stocks of social capital potentially differ in anumber of ways that are difficult to observe. For example, households withhigh social capital may be more entrepreneurial, extroverted or risk-lovingthan other households, and these factors may affect both social capital andincome and generate spurious correlations between our variables of interest.In this respect, VARHS is highly attractive because the panel dimension of thedataset allows us to control for such unobserved household characteristics byincluding household fixed effects (household dummies) in regressions. To theextent that household characteristics do not change systematically over time,they are taken into account by household fixed effects.

Other identification issues are more difficult to solve. Most importantly,causality may in some cases run from income to social capital rather than, orin addition to, running from social capital to income. I cannot rule out thatincome is used as a criterion for Party membership, for example. I cannotfully resolve these issues in this context and therefore it is prudent to viewregressions as ‘descriptive’ rather than ‘structural’. Results are interestingnonetheless.

The different aspects of social capital potentially affect each other in com-plex ways. For example, high levels of trust may increase people’s willingnessto participate in social groups. On the other hand, group participation mayin itself also generate trust. Therefore, it is complicated to separate theeffects of different dimensions of social capital on income from each other.Our approach is to first present regressions where each social capitalmeasure isentered alone, along with the set of exogenous control variables (Table 9.1)and then also estimate models where all variables are entered together(Table 9.2). Table 9.1 presents only fixed effects regression (note that bi-variaterelations between social capital measures and income are shown in the figuresthat present results by income quintile). Table 9.2 presents random as well asfixed effects models. While random effects models do not take account ofunobserved, fixed household characteristics, they allow us to exploit inter-household variation in social capital and other variables and may thereforealso be considered interesting, especially in terms of estimating effects ofvariables that vary little over time, such as ethnicity of the household head.Random effects models include province dummies (not shown).


195

Consider now the results in Table 9.1, where social capital measures areentered one by one (with the exception of the measures of MO and non-MOmembership, which are entered together).Table 9.1 shows positive and significant effects of Party membership, con-

nections with government officials, and informal economic networks (meas-ured by number of potential financial helpers). On the other hand, there areno significant effects of membership in MOs or other groups (in contrast withthe findings on group membership in Narayan and Pritchett 1999). The effectof trust is also just insignificant (Narayan and Pritchett 1999: 143).Next, consider Table 9.2, where all social capital variables are entered together.

Regressions 1 and 3 include random effects, while regressions 2 and 4 arefixed effects models as in Table 9.1. Regressions 3 and 4 include number ofworking-age household members and asset variables along with the controlvariables used in Table 9.1.Communist Party membership is significant and positive in all models. The

estimated return to Party membership is in the order of 10 per cent. This is

Table 9.1 Social capital and income, simple models

Dependent variable: Real income per capita (ln)

FE FE FE FE FE

Party member 0.104***(0.036)

Official (hh member, friend, or relative) 0.043**(0.019)

Number of mass organizations 0.000(0.011)

Number of other voluntary groups 0.004(0.020)

Number of financial helpers 0.009***(0.001)

Trust 0.026(0.018)

Years of schooling, hh head 0.014*** 0.014*** 0.015*** 0.015*** 0.014***(0.005) (0.005) (0.005) (0.005) (0.005)

Age of hh head 0.025** 0.025** 0.026** 0.024** 0.025**(0.010) (0.010) (0.010) (0.011) (0.010)

Age squared/100 �0.028*** �0.028*** �0.028*** �0.027*** �0.028***(0.009) (0.009) (0.009) (0.009) (0.009)

Female hh head 0.123** 0.122** 0.120** 0.122** 0.120**(0.054) (0.054) (0.054) (0.053) (0.054)

Kinh 0.211 0.209 0.207 0.199 0.211(0.149) (0.150) (0.149) (0.147) (0.148)

Year fixed effects Yes Yes Yes Yes YesObservations 8,298 8,298 8,298 8,298 8,298Number of household 2,162 2,162 2,162 2,162 2,162

Note: Standard errors adjusted for commune level clustering. ***p< 0.01, **p< 0.05, *p< 0.1.



196

Table 9.2 Social capital and income, comprehensive models

Dependent variable: Real income per capita (ln)

RE FE RE FE

Party member 0.256*** 0.103*** 0.210*** 0.087**(0.034) (0.035) (0.033) (0.034)

Official (hh member, friend, or relative) 0.067*** 0.028 0.058*** 0.024(0.018) (0.019) (0.018) (0.018)

Number of MOs �0.018* �0.006 �0.016* �0.002(0.010) (0.011) (0.009) (0.011)

Number of other voluntary groups 0.000 �0.005 �0.023 �0.014(0.017) (0.020) (0.017) (0.021)

Number of financial helpers 0.010*** 0.009*** 0.008*** 0.008***(0.001) (0.001) (0.001) (0.001)

Trust 0.025 0.030* 0.031* 0.036**(0.017) (0.018) (0.017) (0.018)

Years of schooling, hh head 0.040*** 0.014*** 0.032*** 0.012**(0.003) (0.005) (0.003) (0.005)

Age of hh head 0.035*** 0.024** 0.041*** 0.026**(0.006) (0.010) (0.006) (0.011)

Age squared/100 �0.030*** �0.026*** �0.036*** �0.029***(0.005) (0.009) (0.005) (0.009)

Female hh head 0.077*** 0.123** 0.069** 0.122**(0.029) (0.053) (0.027) (0.055)

Kinh 0.425*** 0.208 0.331*** 0.219(0.049) (0.146) (0.046) (0.145)

Irrigated land, ln(x+1) �0.002 0.002(0.003) (0.004)

Number of buffaloes �0.013 0.008(0.013) (0.013)

Number of cows �0.021** �0.006(0.010) (0.015)

Number of telephones 0.089*** 0.055***(0.009) (0.009)

Number of motorcycles 0.129*** 0.066***(0.018) (0.015)

Number of bicycles �0.014 �0.006(0.009) (0.005)

Number of pesticide sprayers 0.002 0.022(0.016) (0.018)

Number of cars 0.348*** 0.290***(0.070) (0.076)

Working-age hh members, ln �0.280*** �0.238***(0.029) (0.036)

Year fixed effects Yes Yes Yes YesObservations 8,298 8,298 8,298 8,298Number of household 2,162 2,162 2,162 2,162

Notes: Province dummies included in random effects regressions. Standard errors adjusted for commune level clustering.*** p< 0.01, ** p< 0.05, * p< 0.1.



197

consistent with the findings of a strong correlation between income and Partymembership in Figure 9.1 (panel b). Compared with the figure, the regressionresults allow us to rule out that the correlation is entirely driven by under-lying, unobserved, fixed household characteristics that drive both incomeand Party membership. The results are consistent with the view that Partymembership leads to higher income. They are also consistent with the viewthat the Party uses income as a criterion for membership. Both interpret-ations invite further investigations into the functioning of the CommunistParty at the local level. Compared with Table 9.1, the effect of connectionswith government officials is significant in random but not in fixed effectsmodels. This may indicate that connections with officials proxy for Partymembership in Table 9.1. Alternatively, the effect of connections with offi-cials may operate through Party membership. It is quite conceivable thatpersonal connections with officials, or being an official oneself, eases accessto Party membership.There are no significant, positive effects of MO membership or of member-

ship in other voluntary groups (in fact, the effect of MOmembership is weaklysignificantly negative in random effects models). This means that there is noapparent, private economic return to activities in these groups. This does notrule out that group membership affects other aspects of household welfare, orthat there is a positive, social return to group activities. For example, groupsmay produce public goods (such as provision of information about agricul-tural production techniques) that benefit members and non-members alike.To test for such effects, commune-level analyses may be useful.The effect of informal, economic networks (number of financial helpers)

remains positive in all models. One interpretation is that individuals whomayprovide emergency funding are also useful in other types of economic trans-actions, for example, as trading partners or as providers of credit for invest-ment purposes or working capital.The trust variable is now significant in three out of four models, including

both fixed effects models. High-trust households are estimated to earn about3 per cent higher income per capita than other households. This is a moderateeffect, but it is remarkable nonetheless because it is reasonable to expect thatthe social returns to trust are higher than the private returns (a householdmaybenefit from being trusting because trust induces it to engage in profitable butrisky transactions. However, the partners of these transactions also benefit,leading to a positive externality).Overall, results are consistent with hypotheses of positive, private returns to

bridging social capital (trust), bonding social capital (financial helpers aremost often relatives of the respondent), and political capital (Party member-ship and connections with officials). This supports the notion that socialnetworks and attitudes have important economic effects. These factors cannot


198

be ignored if we seek a comprehensive understanding of the factors behindhousehold welfare and economic development.

I briefly consider the effects of control variables. All models estimate asignificant, positive return to schooling. Note that in fixed effects models,variation in schooling of the household head is mostly driven by changes inthe identity of the head. This is even more so for the gender, age, andethnicity variables. Random effects estimates may therefore be equally ormore interesting than fixed effects estimates for these variables. The esti-mated return to an additional year of schooling is 3–4 per cent in randomeffects models and about 1.4 per cent in fixed effects models. As expected, theeffect of age is inversely U-shaped in all models. In the random effects model,the peak is 57–8 years. The effect of female household headship is signifi-cantly positive in all models, which is somewhat surprising. The explanationmay be that the most common reason for women being household heads iswidowhood. The death of a husband leads to a drop in the denominator ofthe ‘income per capita’ variable. If the husband was old or sick, he may nothave contributed strongly to income generation in recent years, and thecorresponding drop in the numerator resulting from his death is perhapsnot very large.

Ethnicity of the household head varies very little over time and it istherefore not surprising that the effect of being Kinh is insignificant in fixedeffects models. In random effects models, there is a strong (33–43 per cent)and highly significant, positive effect of belonging to the ethnic majority.Since the random effects models include province fixed effects, this effect isnot driven by regional differences. It starkly highlights the disadvantaged,economic position of ethnic minorities (cf. Chapter 13). Among the assetvariables, it is perhaps surprising that land holdings are not significant (thesame is true if total, rather than irrigated, land holdings are entered). Oneplausible interpretation is that there are now a number of other viable liveli-hood strategies than agriculture, even in rural areas, and that focusing onwage labour or non-farm enterprises is often at least as profitable as farming(cf. Ravallion and van de Walle 2008). Among non-land assets, only holdingsof motorcycles and telephones are significant. These variables are potentiallyendogenous and estimates should not be regarded as causal. The effect of thenumber of working-age household members is negative. This implies dimin-ishing, marginal return to labour (the dependent variable being per capitaincome) and indicates the presence of frictions in the labour market.3

3 With perfect labour markets, people can always find work at the going wage rate, implyingconstant returns. On the other hand, if workers are to some extent constrained to working onfamily farms or in other family businesses, diminishing returns are expected.


199

9.8 Conclusions

This chapter has documented the evolution of various aspects of social capitalover time and the distribution of social capital across regions and socioeco-nomic groups. It has also explored the private, economic returns to socialcapital. Results reveal a very strong correlation between Communist Partymembership and household income. Party membership is more common inthe North than in other regions. Membership of MOs is less common in theSouth than in the North but there is no income gradient in MO member-ships. MOs are much more widespread than other voluntary groups, butnon-MOs are growing faster than MOs. This development is driven primarilyby the growth of groups for the elderly. A moderate increase in generalizedtrust was observed between 2008 and 2014. This indicates a strengthening of‘bridging’ social capital in Viet Nam, which is an important prerequisite forcontinued, economic development. While bridging social capital may begrowing, ‘bonding’ social capital also continues to play an important role.In particular, family ties play a very strong role in economic transactionssuch as emergency lending and land rentals. There are no signs that relianceon family ties in economic transactions is declining over time.Income regressions reveal positive effects of political capital, measured by

Communist Party membership and connections with government officials.This is consistent with the view that patronage relations are important inVietnamese politics and highlights the importance of increasing the account-ability of political elites (cf. Appold and Phong 2001; Gillespie 2002;Gainsborough 2007; Markussen and Tarp 2014). There are also positive effectsof informal networks and of generalized trust, indicating the importance of,respectively, bonding and bridging social capital. On the other hand, mem-bership of MOs and other voluntary social groups has no effect on householdincome. This does not rule out that there is a positive, social (community-level) economic return to activities in these groups, or that groups havepositive effects on other aspects of household welfare than income.Future studies should, for example, take further steps to identify the causal

effects of political capital on income, estimate social as well as private returnsto social capital, and further investigate the role of family networks in theVietnamese economy.

Acknowledgements

I am grateful for inputs from Carol Newman, Finn Tarp, Ulrik Beck, and seminarparticipants in Hanoi.


200

References

Alesina, A. E. and E. La Ferrara (2002). ‘Who Trusts Others?’ Journal of Public Economics,85: 207–34.

Appold, S. J. and D. T. Phong (2001). ‘Patron–Client Relationships in a RestructuringEconomy: An Exploration of Interorganizational Linkages in Viet Nam’. EconomicDevelopment and Cultural Change, 50: 47–76.

Dalton, R. J., M. H. Pham, T. N. Pham, and T. O. Ngu-Ngoc (2002). ‘Social Relations andSocial Capital in Viet Nam: Findings from the 2002World Values Survey’. InternationalJournal of Comparative Sociology, 1(3–4): 369–86.

Gainsborough, M. (2007). ‘From Patronage to “Outcomes”: Viet Nam’s CommunistParty Congresses Reconsidered’. Journal of Vietnamese Studies, 2(1): 3–26.

Gillespie, J. (2002). ‘The Political–Legal Culture of Anti-Corruption Reforms in VietNam’. In T. Lindsey and H. Dick (eds), Corruption in Asia: Rethinking the GovernanceParadigm. Leichhardt, NSW: Federation Press.

Gourou, P. (1936). Les Paysans du Delta Tonkinois. Paris: Mouton &Co., andMaison DesScience de L’Homme.

Kinghan, C. and C. Newman (2015). Social Capital, Political Connections and HouseholdEnterprises: Evidence from Viet Nam. WIDERWorking Paper 2015/001. Helsinki: UNU-WIDER.

Knack, S. and P. Keefer (1997). ‘Does Social Capital have an Economic Payoff? A Cross-Country Investigation’. Quarterly Journal of Economics, 112(4): 1251–88.

Markussen, T. and F. Tarp (2014). ‘Political Connections and Land-Related Investmentin Rural Viet Nam’. Journal of Development Economics, 110: 291–302.

Markussen, T., M. Fibæk, N. D. A. Tuan, and F. Tarp (2014). The Happy Farmer: Self-Employment and Subjective Well-Being in Rural Viet Nam. WIDERWorking Paper 2014/108. Helsinki: UNU-WIDER.

Narayan, D. and L. Pritchett (1999). ‘Cents and Sociability: Household Income andSocial Capital in Rural Tanzania’. Economic Development and Cultural Change, 47(4):871–97.

Newman, C. and M. Zhang (2015). Connections and the Allocation of Public Benefits.WIDER Working Paper 2015/031. Helsinki: UNU-WIDER.

Newman, C., F. Tarp, and K. van den Broeck (2014). ‘Social Capital, Networks Effects,and Savings in Rural Viet Nam’. Review of Income and Wealth, 60(1): 79–99.

Popkin, S. L. (1979). The Rational Peasant: The Political Economy of Rural Society in VietNam. Berkeley, CA: University of California Press.

Putnam, R. D. (1993).Making DemocracyWork: Civic Traditions inModern Italy. Princeton,NJ: Princeton University Press.

Ravallion, M. and D. van de Walle (2008). ‘Does Rising Landlessness Signal Success orFailure for Viet Nam’s Agrarian Transition?’ Journal of Development Economics, 87:191–209.

Woolcock, M. and D. Narayan (2000). ‘Social Capital: Implications for DevelopmentTheory, Research and Policy’. World Bank Research Observer, 15(2): 225–49.


201

Part IIIWelfare Outcomes andDistribution Issues

10

Welfare Dynamics:2006–14

Andy McKay and Finn Tarp

10.1 Introduction

A key interest in using household-level panel data is to analyse the evolutionof welfare over time. Among the wide range of information collected in theViet Nam Access to Resources Household Surveys (VARHS) is data whichenables the construction of measures of households’ food consumption,their income from different sources and their ownership of a wide range ofassets. Each of these can form the basis for a welfare measure in its own right.And measures of each of these are available, on a comparable basis, across thefive waves since 2006. Taking advantage of the panel feature of the data setoffers an important opportunity to examine the dynamics of welfare in ruralViet Nam. The fact that three separate measures are available enriches theanalysis and provides cross checks. More specifically, the VARHS makes itpossible to identify cases of consistent progress over the five waves, cases ofregress, and cases of volatility in living conditions. It also contributes tounderstanding the factors which are associated with these changes.

In analytical work based on panel data, it is important to be mindful of thepresence of attrition, an issue already introduced in Chapter 2. It is consideredin what follows specifically in relation to the analysis of the dynamics ofhousehold welfare.

An extensive literature has looked at poverty dynamics based on householdsurvey data. These have included a number of studies for Viet Nam, drawing onthe data sets available from the different rounds of the Viet Nam HouseholdLiving Standards Survey (VHLSS) or the previous Viet Nam Living StandardsSurveys (VLSS). Examples of such studies include Glewwe and Nguyen (2002),Justino, Litchfield, and Pham (2008), Baulch and Dat (2011), Imai, Gaiha, and

Kang (2010), and Coello, Fall, and Suwa-Eisenmann (2010). These studiesreport consistent progress in households escaping from poverty on a largescale over the 1990s and 2000s, evidence which is consistent with existingcross sectional poverty estimates and the general macro developmentsreported in Chapter 1.The focus here is on welfare dynamics without making reference to any

specific poverty line. Quite a lot of literature has examined this question aswell, but mostly in countries other than Viet Nam—examples include Dercon(2004), Fields and colleagues (2003), Jalan and Ravallion (2004), Lokshin andRavallion (2006), Beegle, DeWeerdt, and Dercon (2011), and Hirvonen and deWeerdt (2013). Other studies have sought to model asset dynamics, includingCarter and Barrett (2006), Lybbert and colleagues (2004), and Barrett andcolleagues (2006). The analysis in this chapter draws on insights from thisliterature.This chapter is structured as follows. Section 10.2 explains the construction

of the welfare measures and presents some initial characteristics of the data.In turn, Section 10.3 analyses the structure of income in more detail. Themain welfare dynamics analysis is then presented in Sections 10.4–10.6.Section 10.4 contains a further descriptive analysis, whereas Section 10.5examines the issue of attrition. Section 10.6 presents an econometric analysis,and Section 10.7 offers our conclusions.

10.2 Measuring Household Welfare

As noted in Section 10.1, there are three different ways of assessing thewelfare levels of households based on the VARHS data: food consumption,income, and assets. Information is collected on household consumption ofmain food commodities over the preceding four weeks (from purchases, ownproduction, or other sources). In all five waves households were asked abouttheir summary income from different main sources (agriculture, wage, non-farm non-wage, transfers, etc.). It is, in addition, possible to construct othermeasures of household income from the VARHS data, using detailed ques-tions about agricultural sales and input purchases, on members’ engagementin wage work, on household non-wage non-farm activities, and on commonproperty resources, as well as receipts of transfers. But as not all of thesecomponents were collected in the 2006 survey, this measure can only becomputed for the last four waves. Themain focus here is onmeasures availablethroughout the period considered, but this second income measure is brieflyreferred to as well.Both the food consumption and income measures were expressed on a per

capita basis, and then adjusted for price differences over time and between the

Welfare Outcomes and Distribution Issues

206

different provinces in Viet Nam. The price adjustment over time for theincome measure is made using the rural value of the consumer price index(CPI) for Viet Nam at the province level; and the adjustment over time for thefood consumption measure is made using the province-level value of the foodprice index from the consumer price index. Both indices were supplied by theGeneral Statistics Office of Viet Nam (GSO). The spatial adjustment is madefrom the eight-regional spatial price index for Viet Nam computed from the2010 round of the VHLSS.

In the case of household assets, information on ownership of a wide rangeof different types is available in the data set. A summary measure is needed.Asset prices may not be reliable, and some assets may not be easily valued atall, so we construct a summary measure here using factor analysis followingthe principles set out by Sahn and Stifel (2000). The asset index is constructedto include: land and productive assets owned by the household; consumerdurable goods; human capital; andmeasures of social capital. The precise formof the index is presented in Table 10.A1 in the Appendix.

Summary statistics for the different welfare measures for the householdsincluded in the panel for the five waves are presented in Table 10.1 and kerneldensity plots for the same variables are presented in Figure 10.1. Both themean and median values of the real food consumption measures increasebetween each wave and the next. The median value of the asset index alsoincreases consistently over this period and the mean value increases in mostperiods. The kernel densities show a general pattern of rightward shifts,though not always across the whole distribution.

Turning to income, for the years for which both measures are available, theestimates are of similar orders of magnitude. Using the summary incomemeasure, the mean and median values both increase consistently between2006 and 2010, and particularly strongly between 2008 and 2010. They then

Table 10.1 Summary properties of VARHS welfare measures

Variable 2006 2008 2010 2012 2014

Asset index Mean 0.042 0.040 0.198 0.320 0.346Standard deviation 1.068 1.084 1.092 1.082 1.082Median �0.032 �0.005 0.136 0.290 0.327

Food consumption Mean 265.9 284.9 319.7 411.6 416.1Standard deviation 276.3 274.0 214.6 293.8 291.3Median 200.0 223.8 268.0 339.8 345.7

Household income 1 Mean 13,068 16,983 28,517 24,090 27,376Standard deviation 19,224 25,447 42,743 31,095 37,680Median 8,614 10,755 18,822 16,860 18,954

Household income 2 Mean 15,483 19,848 22,738 25,051Standard deviation 19,681 30,600 27,025 26,968Median 10,977 13,976 16,860 18,789

Source: Authors’ computation from VARHS panel database.

Welfare Dynamics: 2006–14

207

c: Asset index

d: More detailed income measure

0 5

0.40.30.20.10.0

0.40.30.20.10.0

0.40.30.20.10.0

0.40.30.20.10.0

0.40.30.20.10.0

0.60.40.20.0

0.60.40.20.0

0.60.40.20.0

0.60.40.20.0

0.60.40.20.0

0.40.20.0

0.40.20.0

0.40.20.0

0.40.20.0

0.40.20.0

0.4

0.2

0.0

0.4

0.2

0.0

0.4

0.2

0.0

0.40.20.0

Den

sity

Den

sity

Den

sity

Den

sity

a: Food consumption

b: Summary income measure

0.0 2.5 5.0 7.5 10.0

6 9 12 6 9 12

Figure 10.1 Kernel density plots of different welfare measuresSource: Authors’ illustration based on data from VARHS database. In each case the oldest year is at the top of the panel.

fall between 2010 and 2012 before rising again in 2014. The kernel densityplots show a pattern of shifts to the right over time, again demonstrating thelarge increase in the summary income measure between 2008 and 2010. Themore detailed income measure shows a rather smaller increase between 2008and 2010. It may be that the summary measure was less accurately reported in2010 than in other years and the detailed one is more accurate (which cannotbe known for sure). But it is also likely that inflation over this period, wherefood prices increased sharply, has been underestimated because the CPI placestoo little weight on food for these rural households. This is not an issue withthe food consumption measure.

In all cases though, the overall pattern of the three measures available forthe full period is of the welfare levels being much higher in 2014 than was thecase in 2006.

For the most part, the different welfare measures show very similar patternsof change over time; the only notable difference being between the summaryincome measure and the other two. It is, however, fully expected that incomewill be somewhat more volatile over time, reflecting short term factors, thanthe food consumption or asset measures. Thus, this difference is not surpris-ing, reinforced by the fact that income is also likely to be less accuratelymeasured than the other two measures.

10.3 Sources of Income

Before turning to analysing welfare dynamics in detail, we first considersources from where households derive their incomes. The summary measureof total household income was estimated as a sum of reported income fromeach of several main sources; and the percentage composition of total incomefor the five years of the panel is reported in Table 10.2.

The two most important sources of income are agricultural income andwage income. Between 2006 and 2014 the relative importance of wageincome gradually increased and that of agricultural income fell, but bothsources remained important over the full period. The vast majority of house-holds earn income from agriculture. For instance, 90 per cent of the panelhouseholds earned at least some income from agriculture. While this pro-portion was higher in 2006 and 2008 and fell slightly over time, even by2014 nearly 84 per cent of the panel households reported some income fromagriculture.

All other income sources are earned by a smaller proportion of house-holds. Some 65 per cent of households reported earning wage income andfor the remaining income sources the proportions were smaller. While


209

agriculture therefore continues to play a central role in the livelihood ofmost of these rural households in Viet Nam, almost all households com-bine agriculture with other income sources. Fewer than 4 per cent ofhouseholds earned their livelihood from agriculture only in 2008 and thisproportion was lower in all other years. Wage income and the other earn-ings categories (much of which is accounted for by transfers) are the nextmost important sources for most households. Smaller amounts are earnedon average from a household business activity or from common propertyresources.Returning to the amounts of earnings, consistentlymore than 60 per cent of

earnings come from agriculture and wages combined. These results reconfirmthe fact that over time the share of wages increases and that of agriculture falls.The increase in wage income reflects both the fact that over time morehouseholds have a member engaged in wage work as well as an increase inearnings. Only a minority of households have businesses. Yet, for many thatdo this is a good income source. While many more households receive trans-fers the amounts are typically smaller. The other components of incomereported in Table 10.2 are very small on average, though they can be import-ant for some individual households.It is important as well to note that the composition of income is relatively

stable over time, providing some support for the view that the large increase in

Table 10.2 Composition of household income using summary incomemeasure (%)

2006 2008 2010 2012 2014

Wage 28.6 28.2 29.7 33.2 35.7Agriculture 36.9 41.3 36.7 30.0 28.4Common property resources 2.7 3.4 3.2 2.9 2.1Non-farm non-wage 13.9 12.3 11.8 11.8 12.1Rental 0.5 0.7 0.8 1.0 0.7Other 17.3 14.0 17.8 21.2 21.0

Six mainly agricultural provincesWage 22.9 14.8 18.3 21.8 26.0Agriculture 54.4 62.6 57.5 53.6 50.0Common property resources 4.3 6.3 4.8 5.3 3.9Non-farm non-wage 8.8 7.2 8.4 6.5 6.4Rental 0.4 0.3 0.5 0.7 0.3Other 9.3 8.8 10.4 12.0 13.5

Six less agricultural provincesWage 30.7 33.1 33.9 37.4 39.2Agriculture 30.5 33.5 29.1 21.3 20.5Common property resources 2.1 2.4 2.5 2.0 1.4Non-farm non-wage 15.7 14.2 13.1 13.7 14.2Rental 0.6 0.8 0.9 1.1 0.9Other 20.3 16.0 20.5 24.5 23.8



210

this measure between 2008 and 2010 may have reflected an underestimationof inflation.

The lower panel of Table 10.2 shows a disaggregation between the six ofthe twelve VARHS provinces where agricultural livelihoods remain veryimportant, and the other six where non-agricultural livelihoods are becom-ing increasingly important over time.1 The former provinces are those in thehighlands in the north of the country and the Central Highlands; whilethe latter set groups together provinces close to major cities and/or locatedon the coast. Table 10.2 shows important differences between these groups.In the former group, agriculture accounts for the majority of income onaverage, though its share has been slowly declining since 2008. Wageincome only accounts for around one-fifth to one-quarter of income, withbusiness income accounting for another 10–15 per cent. The share of otherincome is relatively small. The share of wage income has been increasingfrom 2008 but remains much less important than agriculture. In the secondgroup of provinces, wage income is the biggest source throughout, and it isgrowing over time. However, agriculture continues to make a significantcontribution even if much less than in the former group. In the six lessagricultural provinces both business income and income from transfers aremore important than in the former group of provinces.

In addition to the summary measures of income reported to date, the lastfour rounds of the VARHS collected more detailed information enabling thecomputation of more precise and more disaggregated components of house-hold income. As well as providing more detail, this may be a more accurateestimate of household income. Table 10.3 reports mean values of per capitareal income from the more detailed income measure defined in this way.

1 The first six provinces are: Lao Cai, Lai Chau, Dien Bien, Dak Lak, Dak Nong, and Lam Dong.The second six are Ha Tay, Phu Tho, Nghe An, Quang Nam, Khanh Hoa, and Long An.

Table 10.3 More detailed analysis of household income composition,2008–14 (%)

2008 2010 2012 2014

Crops 34.6 23.2 22.9 23.6Livestock 4.6 3.6 8.0 �7.6Common property resources 3.4 4.4 2.9 2.1Household business 12.7 13.7 3.9 12.4Wages 28.2 31.3 33.0 44.5Private transfers 8.0 12.1 10.5 10.7Public transfers 6.8 8.2 8.6 9.9Other 1.7 3.5 10.4 4.4Per capita household income 15,483 19,848 22,738 25,051



211

The broad consistency in the levels of the summary estimates seen in theSection 10.2 is quite similar to the more detailed measures. The overall sharesof the main income components are also quite similar. The greater detailwhich is available here shows that private transfers seem to be the mostimportant form of other income as defined in Table 10.2, and among agricul-tural earnings from crops dominate with livestock and aquaculture makingmuch smaller contributions on average.We now turn to the principal interest in this chapter; that is changes in

household welfare over time.

10.4 Descriptive Analysis of Changes in Household Welfare

Section 10.2 reported the aggregate change in the three welfare measuresconsidered here. Given the likelihood of diversity of experience within thepanel, a much more disaggregated analysis is also called for. This is thefocus of the remainder of this chapter. We focus on the measures availablefor all the five waves; and begin by summarizing the patterns and trendsin real per capita food expenditure consumption among the households.This is reported in Table 10.4 disaggregated according to different criteria,some of which will be considered again in the multivariate analysis ofSection 10.6.The average level of food expenditure is seen to be significantly lower in the

provinces of the North East and NorthWest (Lao Cai, Lai Chau, and Dien Bien)than anywhere else. Focusing on changes, across the sample there is largeaverage growth of food expenditure at an annualized average of 5.7 per cent.Figure 10.1 showed that the 2014 distribution clearly lies to the right of the2006 distribution. This suggests growth across the distribution, but does notimply that consumption grew for all individual households. Among mostprovinces, average levels of consumption tend to fluctuate from one year toanother, and the fastest growth over the 2006–14 period was experienced inHa Tay, Quang Nam, and Long An, all provinces located close to importanturban centres.The panel data makes it possible to study mobility over time. Given inevit-

able measurement error in the data, it makes sense to focus on larger changes.Accordingly, Table 10.5 reports the percentage of households experiencingeither: (i) a 20 per cent or greater increase in real per capita food consumptionor income between 2006 and 2014, or (ii) a 20 per cent or larger reduction,disaggregated according to the same categories as in Table 10.4. Overall, nearly67 per cent of households experienced increases of 20 per cent or more in theirfood consumption over the period, and nearly 76 per cent an increase in theirincome. So a majority of households in all categories saw increases of 20 per


212

cent ormore in both variables over the period, though in some instances, suchas the case of Lao Cai in the Northern Highlands, only just over half of thehouseholds saw increases of this magnitude in their food consumption orincome.

What is also striking is that 17.6 per cent of households, more than one inevery six, saw their food consumption levels falling by 20 per cent or morebetween 2006 and 2014 even when the average increase was 5.7 per cent a year

Table 10.4 Levels and changes in real per capita food consumption in the 2006–14VARHS panel

2006 2008 2010 2012 2014 Annualizedgrowth rate2006–14, %

By provinceHa Tay 247.5 288.8 347.5 488.3 477.2 8.6Lao Cai 235.0 166.0 140.4 215.5 286.7 2.5Phu Tho 297.3 330.5 380.5 475.2 365.0 2.6Lai Chau 163.1 187.7 177.2 231.8 198.7 2.5Dien Bien 195.5 187.9 269.2 275.7 285.2 4.8Nghe An 236.1 304.2 269.1 395.7 316.6 3.7Quang Nam 259.1 305.2 315.8 403.3 498.7 8.5Khanh Hoa 362.9 270.3 467.3 385.0 509.3 4.3Dak Lak 276.7 289.2 252.4 347.6 371.5 3.8Dak Nong 346.8 351.1 337.7 446.7 387.4 1.4Lam Dong 350.8 206.2 330.9 321.4 432.8 2.7Long An 287.8 303.0 359.6 458.4 526.5 7.8

By education quartileLowest 177.2 194.6 221.1 278.4 296.2 6.62 237.2 247.1 299.4 382.5 400.4 6.83 286.1 302.3 357.0 430.8 457.4 6.0Highest 370.5 405.5 404.9 566.1 514.2 4.2

By household size1 or 2 288.0 311.7 349.3 450.6 455.5 5.93 or 4 173.6 173.5 195.7 250.7 253.4 4.85 or 6 288.0 311.7 349.3 450.6 455.5 5.9More than 6 173.6 173.5 195.7 250.7 253.4 4.8

By ethnicityKinh 288.0 311.7 349.3 450.6 455.5 5.9Non-Kinh 173.6 173.5 195.7 250.7 253.4 4.8

By remoteness statusNon-remote 282.4 295.9 337.8 430.0 426.3 5.3Remote 227.5 259.2 277.5 369.0 392.3 7.1

By illness statusNot ill 273.4 299.3 323.8 434.4 425.1 5.7Suffered illness 253.9 261.9 313.1 375.7 401.8 5.9

By migrant statusNo migrants 262.0 278.6 323.3 393.9 388.2 5.0Migrants 272.8 296.1 313.1 443.4 466.0 6.9Total 265.9 284.9 319.7 411.6 416.1 5.8

Note: Disaggregated by different criteria (VND ‘000s).



213

over eight years. A non-negligible number of households have been gettingsignificantly worse off, while around them many households improved theirliving conditions substantially. This is much more common in some provinces(often Northern or Central Highlands) and is higher among ethnic minorityhouseholds (Table 10.5). The results for household income also show 11.6 percent of households experiencing sizeable reductions over this period. Theseproportions are again significantly higher among ethnic minority households

Table 10.5 Percentage of households experiencing significant increases andreductions in food expenditure and income over the period of the surveys

Food consumption Income

% increasing20% or more

% falling 20%or more

% increasing20% or more

% falling 20%or more

By provinceHa Tay 77.0 9.8 79.1 10.0Lao Cai 54.1 21.2 55.3 32.9Phu Tho 57.6 23.2 75.8 9.4Lai Chau 61.7 26.2 72.9 12.2Dien Bien 63.6 25.3 68.7 16.2Nghe An 62.2 21.3 76.6 11.7Quang Nam 77.3 6.8 84.9 5.4Khanh Hoa 57.8 12.7 80.6 4.2Dak Lak 57.7 26.9 65.7 17.6Dak Nong 53.9 33.0 73.9 17.4Lam Dong 53.1 31.3 67.2 18.8Long An 72.8 14.9 76.5 9.8

By education quartileLowest 68.0 17.3 73.8 12.92 69.3 16.2 74.3 12.23 66.8 17.1 79.2 9.5Highest 62.6 20.4 75.5 11.9

By household size1 or 2 61.4 22.8 62.2 16.33 or 4 63.3 18.6 76.0 12.95 or 6 72.0 14.9 78.6 9.0More than 6 67.2 18.1 78.6 10.9

By ethnicityKinh 67.6 16.2 76.6 10.6Non-Kinh 63.7 23.6 72.3 15.8

By remoteness statusNon-remote 66.1 18.2 76.1 11.7Remote 68.4 16.3 75.0 11.3

By illness statusNot ill 66.7 17.5 76.8 11.4Suffered illness 67.0 17.8 74.2 12.0

By migrant statusNo migrants 64.4 18.9 72.0 13.4Migrants 71.1 15.4 82.5 8.3Total 66.8 17.6 75.8 11.6



214

and also in Lao Cai province compared to other groups. For neither measure isthere any reason to think that this just reflects life cycle factors as householdsage or children leave home; while this might be the explanation in some casesan analysis of the data suggests that there are many other factors.

10.5 Attrition

The issue of attrition was introduced in Chapter 2, where it was noted that thescale of attrition is small in the VARHS data and does not show any systematicpattern according to the variables considered there. In coming to grips withhousehold welfare dynamics, it is important to consider here whether there isany systematic relationship between attrition and the welfare measures used.

A comparison between the average values of the different welfare measuresamong attrited households and those remaining in the panel is presented inTable 10.6. The number of attrited households is very similar to those alreadyreported in Chapter 2 (any differences being accounted for by the fact thatthese tables are only computed for the income, consumption and asset meas-ures in focus).

There is no significant difference in the baseline food expenditure or incomein any of the sub-panels presented. There are, however, significant differencesbetween attrited and retained households in relation to their asset holdings,

Table 10.6 Extent and nature of attrition in the VARHS 2006–12 panel

Samplesize

Numberattrited

Mean:attrited

Mean: non-attrited

t test forNA-A=0

Summary income2006 base 23222006-8 panel 2264 58 13,234 13,125 �0.042006-8-10 panel 2223 41 14,102 13,107 �0.332006-8-10-12 panel 2185 38 12,096 13,124 0.332006-8-10-12-14 panel 2160 25 18,012 13,068 �1.28

Food consumption2006 base 23222006-8 panel 2264 58 279.8 266.8 �0.362006-8-10 panel 2223 41 316.1 265.9 �1.162006-8-10-12 panel 2185 38 284.4 266.2 0.42006-8-10-12-14 panel 2160 25 291.8 265.9 �0.47

Asset index2006 base 23242006-8 panel 2266 58 �0.628 0.184 4.532006-8-10 panel 2225 41 �0.376 0.026 2.382006-8-10-12 panel 2187 38 �0.595 0.036 3.612006-8-10-12-14 panel 2162 25 �0.464 0.042 2.36



215

with, in every round, attrited households having less assets than those thatremain. Having fewer assets may relate to the life cycle stage for the household.It does seem that those with fewer assets are more likely to migrate. Overallthough, there is only very limited evidence of systematic patterns of attrition.

10.6 Econometric Analysis of Welfare Change

We now move on to conduct a multivariate analysis of welfare change in thepanel to properly identify the factors associated with positive and negativechanges. In so doing, we consider two different approaches: examiningchanges between the beginning and end of the panel for households presentin all five waves, then looking at wave-to-wave changes for all householdspresent for the two waves in question, within the balanced panel.What is being estimated here is effectively a growth model at the micro

level, where the change in the logarithm of the welfare measures is regressedon their respective level in the previous period and different household char-acteristics in the previous period, including fixed effects (variously done at theprovince and district level). In this model the previous period value of thewelfare measure is highly likely to be endogenous. Instrumental variables aretherefore needed, and for both income and consumption various physicalassets owned by the household are relied on. In the case of assets, the issueof endogeneity of the level of the base previous period asset values is perhapsless a matter of concern. In addition it is difficult to identify an instrumentalvariable for this variable; so this model is simply estimated by ordinary leastsquares (OLS).Table 10.7 presents values for the change in the welfare measures between

the beginning and end of the period, while Table 10.8 shows the wave onwave changes within the panel. All these models are estimated with district-level fixed effects. In the cases of food consumption and income, the baseperiod levels of these variables are clearly shown to be endogenous accordingto the Wu–Hausman test. Household ownership of motorcycles and tele-phones in the base period clearly function as strongly significant instrumentalvariables in each case. The first stage F statistics are comfortably above thestandard thresholds and there is no evidence of over-identification.In all cases the lagged level of the welfare measure is significant and nega-

tive, as expected in a growth model. Beginning with the regressions whichcompare the welfare outcomes at the start to those at the end, education isstrongly significant in relation to household income and assets, though sur-prisingly not in the case of food consumption. The fact of a household havinghad migrants has a large and strongly significant positive influence on all


216

three measures of welfare; though in these models the fact of being from anethnic minority is only significantly negative in the income model. Thisvariable is also significant in relation to food consumption in an equivalentmodel including province level fixed effects; the district-level fixed effectshave made this variable insignificant in this model.

Overall, household size has an unsurprisingly significant negative associ-ation with changes in per capita food consumption and income. Householdcomposition variables are often significant in these models, with householdshaving more members in the 15–60 and sometimes 5–15 range often having asignificant positive influence on changes in welfare. Time spent working inthe base period is negatively associated with income growth, and a naturalshock experienced in the base period is positively associated with asset accu-mulation, perhaps as a subsequent reaction to this shock.

Table 10.7 Regression results for changes in welfare measures from 2006–14 (with district-level fixed effects)

Food consumption Income Asset index

Coef. z Coef. z Coef. t

Food cons, 2006 �0.5381 �7.83Income 2006 �0.3747 �5.04Asset index 2006 �0.6330 �23.57Time worked �0.0004 �5.13 0.0000 �0.35Hh size �0.0687 �1.81 �0.0726 �1.62 �0.0777 �1.38Females < 5 years �0.0020 �0.04 0.0802 1.24 0.0688 0.89Males < 5 years �0.0137 �0.27 0.0870 1.39 0.1135 1.52Females 5–15 years 0.0413 1.02 0.1233 2.44 0.2263 3.81Males 5–15 years 0.1001 2.47 0.1345 2.70 0.3079 5.18Females 15–59 years 0.0670 1.66 0.1043 2.09 0.0739 1.24Males 15–59 years 0.0558 1.48 0.1548 3.34 0.1811 3.22Females 60 years and above 0.0506 0.86 0.1017 1.44 0.0644 0.74Education per capita 0.0057 0.82 0.0150 1.88 0.0606 7.66If household has business �0.0128 �0.41 �0.0057 �0.15 �0.0106 �0.24If had natural shock �0.0825 �1.39 0.0550 0.78 0.1352 1.54If had pest attack 0.0275 0.73 0.0598 1.30 0.0079 0.14If had economic shock 0.1092 0.56 0.4413 1.92 0.1909 0.67If had illness shock �0.0381 �0.97 �0.0654 �1.41 �0.0611 �1.07Number of groups �0.0441 �1.65 �0.0287 �0.92Number of political groups 0.0442 1.49 0.0386 1.10If female-headed �0.0190 �0.51 �0.0330 �0.74 �0.2130 �3.86If has red book �0.0439 �0.97 �0.0471 �0.89 0.1084 1.64If remote 0.0208 0.62 �0.0288 �0.72 �0.0300 �0.60If from ethnic minority �0.0816 �0.97 �0.1644 �1.66 �0.0396 �0.32Minority*education 0.0048 0.42 0.0069 0.50 0.0014 0.08If have absent household

member0.1709 5.83 0.1917 5.40 0.0926 2.13

Constant 3.4430 8.31 4.0217 5.58 0.1190 0.33F stat (first stage) 56.5 64.5R square 0.563 0.459 0.379Number of observations 2153 2148 2153

Source: Authors’ estimation based on VARHS panel database.


217

Perhaps of greater interest are the results relating to the entire panel data set(Table 10.8). Again, the instrumental variables strongly pass the test for weakinstruments. Similar results are observed here to what was seen in the previouseconometric model that compared 2006 and 2014, but there are also signifi-cant differences. Education is significantly positively associated with welfarechange in all three models. Again some household composition variables arerelevant. Household size has a positive association with asset accumulation(this variable is measured at the household level), and having more youngboys in the households is negatively associated with the growth in foodconsumption; having more younger people in the household also tends tobe associated with reduced asset accumulation.The household head being from a minority group is now associated with a

large negative influence on food consumption and quite a large negative effect

Table 10.8 Regression results for changes in welfare measures within the VARHS panel(with district-level fixed effects)

Food consumption Income Asset index

Coef. z Coef. z Coef. t

Food cons, t-2 �0.3314 �6.72Income, t-2 �0.2375 �5.72Asset index, t-2 �0.6165 �54.49Time worked �0.0001 �2.49 �0.0003 �6.06 0.0001 3.67Hh size �0.0085 �0.37 �0.0099 �0.42 0.0752 2.99Females < 5 years �0.0143 �0.45 0.0449 1.31 �0.1063 �2.90Males < 5 years �0.0831 �2.60 �0.0054 �0.16 �0.1099 �3.01Females 5–15 years �0.0265 �1.11 0.0016 0.06 �0.0369 �1.35Males 5–15 years �0.0119 �0.49 0.0109 0.42 �0.0291 �1.06Females 15–59 years 0.0219 0.94 0.0532 2.10 0.0191 0.72Males 15–59 years 0.0254 1.17 0.0632 2.71 0.0516 2.1Females 60 and above 0.0310 0.92 0.0304 0.85 �0.0263 �0.68Education per capita 0.0091 1.94 0.0167 3.61 0.0644 16.09If household has business 0.0135 0.71 �0.0487 �2.36 �0.0228 �1.15If had natural shock �0.0044 �0.22 0.0626 2.84 �0.0113 �0.47If had pest attack 0.0345 1.89 0.0217 1.11 �0.0495 �2.36If had economic shock �0.0030 �0.08 �0.0048 �0.12 �0.1060 �2.52If had illness shock �0.0339 �1.47 �0.0641 �2.62 �0.0706 �2.64Number of groups �0.0432 �3.68 0.0276 2.30Number of political groups 0.0303 2.36 �0.0293 �2.15If female-headed �0.0358 �1.68 �0.0365 �1.62 �0.1930 �7.82If has red book �0.0572 �2.35 0.0179 0.72 0.0482 1.77If remote �0.0202 �1.09 �0.0276 �1.40 0.0068 0.31If from ethnic minority �0.1505 �2.78 �0.1009 �1.77 �0.0474 �0.76Minority*education 0.0149 2.26 0.0030 0.42 0.0064 0.83If have absent household member 0.0257 1.49 0.0257 1.4 0.0910 4.56Constant 1.9853 6.88 2.3974 5.81 �0.5405 �3.26F stat (first stage) 167.8 237.3R square 0.324 0.338 0.287Number of observations 8526 8511 8540

Source: Authors’ estimation based on VARHS panel database.


218

on income, despite the presence of district-level fixed effects in the model. Butin the case of food expenditure this effect is increasingly offset as the leveleducation increases. However, the association with migrants is small here (interms of the coefficient) and less significant, except in relation to assets wherethere is a large positive effect. Female-headed households have significantlylower levels of increase in all three welfare measures, with the effects of thisbeing particularly large in relation to assets. Negative shocks experienced inthe previous period have a negative impact on asset accumulation. This modelobviously captures shorter term influences on wellbeing.

Clearly the models used here bring out the beneficial effects of educationand of the presence of migrants in the households as strong positive influ-ences of improvements in wellbeing, with the former effect being stronger inthe short term and the latter being stronger in the longer term. It also clearlyhighlights the disadvantages of being from an ethnic minority (or from adistrict where ethnic minorities are concentrated), as well as the short termdisadvantages faced by female-headed households. Some of these results wereapparent from the descriptive analysis, though others were not.

10.7 Conclusions

The aim of VARHS has, throughout, been on documenting the wellbeing ofrural households focusing, in particular, on access to and the use of productiveresources. Many of the characteristics of the rural households surveyed over theperiod 2006 to 2014 do not change over time, as one would expect, given thatthe same households are surveyed in each year. Nevertheless, some notabledifferences exist. The number of surveyed households classified as poor by theMinistry of Labour, Invalids and Social Affairs (MOLISA) has declined. Thissuggests that, overall, living conditions have in general improved for the sur-veyed households. This is confirmed in this study based on three measures ofwelfare: (i) food consumption, (ii) household income, and (iii) householdownership of assets. These three measures all bear witness to the considerableprogress that has taken place in Viet Nam in the period under study.

However, this is not consistently the case across all areas of the country.Moreover, the welfare measures often show quite a lot of volatility from onesurvey to another, even in indicators such as food expenditure and assets thatshould be thought to be quite stable. One striking finding from the analysis ofthe welfare measures is the failure of Lao Cai to make significant progress overthis period, a period over which most provinces, including some initiallypoorer ones from the north-west, advanced significantly. This is true through-out each of the two year sub-periods as well. It is clearly important to seek to


219

understand the factors which have contributed to a failure of progress in LaoCai over this period.The data also shows that even in provinces where average living conditions

improved a lot, the situation deteriorated for a substantial minority of house-holds in almost every case. Thus, while the aggregate story confirms the picturesfrom VHLSS surveys and elsewhere of significant poverty reduction in ruralViet Nam, the analysis presented here confirms that for a lot of households thesituation has clearly worsened over this period. It is important to understandthis diversity of experience, and the multivariate analysis provides insights.Having a sufficient level of education is associated with a greater likelihood ofbecoming better off, as does having more prime-age household members (andfewer dependents). Migration of some householdmembers also appears to havea very positive impact on the remaining household members over the longerterm; but being of non-Kinh ethnicity is significantly associated with smallerincreases in food consumption and income.The ethnic differential story is well known in Viet Nam. It has also been the

subject of many high-profile policy interventions. The results in this chaptersuggest strikingly that even today being of a non-Kinh ethnicity remains asubstantial disadvantage. The key policymessage emerging is that whilemuchhas been achieved in Viet Nam in terms of growth and poverty reduction,important challenges remain to ensure inclusive progress in the years to come.

Table 10.A1 Factor index weights for asset index

Variable Weight

Years of education per capita 0.171Number of active household members 0.105Number of plots owned 0.051Total area owned 0.035Irrigated area owned 0.049Number of cows 0.039Number of buffalos 0.000Number of pigs 0.024Number of chickens 0.027If household has a business 0.032Number of colour TVs 0.074Number of videos/DVDs 0.074Number of telephones 0.061Number of motorcycles 0.094Number of bicycles 0.079Number of pesticide sprayers 0.041Number of cars 0.034Number of groups attended 0.391Number of political groups 0.407Area of dwelling 0.054If has a good lighting source 0.050If has a toilet 0.067If has a good drinking water source 0.042

Source: Authors’ computation from the VARHS database.


220

References

Barrett, C. B., P. P.Marenya, J.McPeak, B.Minten, F.Murithi,W. Oluoch-Kosura, F. Place,J.-C. Randrianarisoa, J. Rasambainarivo, and J. Wangila (2006). ‘Welfare Dynamics inRural Kenya and Madagascar’. Journal of Development Studies, 42(2): 248–77.

Baulch, B. and V. Hoang Dat (2011). ‘Poverty Dynamics in Vietnam, 2002 to 2006’. InB. Baulch (ed.),Why Poverty Persists: Poverty Dynamics in Asia and Africa. Cheltenham:Edward Elgar.

Beegle, K., J. De Weerdt, and S. Dercon (2011). ‘Migration and Economic Mobility inTanzania: Evidence from a Tracking Survey’. Review of Economics and Statistics, 93(3):1010–33.

Carter, M. R. and C. B. Barrett (2006). ‘The Economics of Poverty Traps and PersistentPoverty: An Asset-Based Approach’. Journal of Development Studies, 42(2): 178–99.

Coello, B., M. Fall, and A. Suwa-Eisenmann (2010). Trade Liberalization and PovertyDynamics in Vietnam, 2002 to 2006. Paris School of Economics Working Paper No.2010–13. Paris: Paris School of Economics.

Dercon, S. (2004). ‘Growth and Shocks: Evidence from Rural Ethiopia’. Journal ofDevelopment Economics, 74(2): 309–29.

Fields, G. S., P. L. Cichello, S. Freije, M. Menéndez, and D. Newhouse (2003). ‘HouseholdIncome Dynamics: A Four-Country Story’. Journal of Development Studies, 40(2): 30–54.

Glewwe, P. and P. Nguyen (2002). Economic Mobility in Vietnam in the 1990s. WorldBank Policy Research Working Paper No. 2838. Washington DC: World Bank.

Hirvonen, K. and J. de Weerdt (2013). Risk Sharing and Internal Migration. World BankPolicy Research Working Paper No. 6429. Washington DC: World Bank.

Imai, K. S., R. Gaiha, and W. Kang (2010). ‘Vulnerability and Poverty Dynamics inVietnam’. Applied Economics, 43(25): 3603–18.

Jalan, J. and M. Ravallion (2004). ‘Household Income Dynamics in Rural China’. InS. Dercon (ed.), Insurance against Poverty. Oxford: Oxford University Press.

Justino, P., J. Litchfield, and H. T. Pham (2008). ‘Poverty Dynamics during TradeReform: Evidence from Rural Vietnam’. Review of Income and Wealth, 54(2): 166–92.

Lokshin, M. and M. Ravallion (2006). ‘Household Income Dynamics in Two TransitionEconomies’. Studies in Nonlinear Dynamics & Econometrics, 8(3): 1182.

Lybbert, T. J., C. B. Barrett, S. Desta, and D. L. Coppock (2004). ‘Stochastic WealthDynamics and RiskManagement among a Poor Population’. Economic Journal, 114(498):750–77.

Sahn, D. E. and D. Stifel (2000). ‘Poverty Comparisons over Time and across Countriesin Africa’. World Development, 28(12): 2123–55.


221

11

Gender Inequality and the Empowermentof Women

Carol Newman

11.1 Introduction

Over the last two decades, a number of changes have been made to Vietnam-ese law to improve the rights and economic situation of women. The 2003Land Law allowed for the joint titling of land, which primarily affectedwomen in allowing them to be named on their husband’s land title. Thegender equality law implemented in Viet Nam in 2006 aimed to ensureequal rights of women in all aspects of economic and political life. Thesechanges were partly driven by efforts to attain Goal 3 of the MillenniumDevelopment Goals (MDGs), which was to ‘Promote Women and EmpowerWomen’.1 With the end of the timeframe for completion of the MDGs uponus, examining gender disparities and how they have evolved over the lastdecade is timely.Other studies have found that the economic situation of women in Viet Nam

has improved, but that gaps still remain. In 2011, for example, the WorldBank Viet Nam Country Gender Assessment pointed to significant progress inrelation to poverty and wellbeing, employment and livelihoods, and politicalparticipation (World Bank 2011). This report highlighted a number of genderdifferences that still remained, including wage disparities (although muchimproved), the over-representation of women in more vulnerable jobs, vul-nerability of older women, particularly in rural areas, and a lack of voice

1 See the ‘Introductory Statement on Vietnam’s Combined 5th and 6th National Report on theImplementation of the UN Convention on the Elimination of All Forms of Discrimination againstWomen (CEDAW)’ by Mme Ha Thi Khiet: <http://www.un.org/womenwatch/daw/cedaw/cedaw37/statements/delegations/VIETNAM.pdf> (accessed 4 April 2016).

http://www.un.org/womenwatch/daw/cedaw/cedaw37/statements/delegations/VIETNAM.pdf

http://www.un.org/womenwatch/daw/cedaw/cedaw37/statements/delegations/VIETNAM.pdf

among women in public positions. More recently, the World Bank’s (2016)report Vietnam 2035, also emphasized that progress in relation to genderequality has been made but that there are still differences along gender lines.In this chapter we focus on gender differences in rural Viet Nam for the period2008 to 2014, highlighting in particular, the heterogeneity in gender inequal-ity for different groups of households and cohorts of women and men.

A large and growing literature emphasizes the value of gender inequality,not only as an end-goal in itself but also in terms of its impact on economicdevelopment. For example, Jensen (2012) and Heath and Mobarak (2015)provide evidence that increasing labour market opportunities for women inIndia and Bangladesh, respectively, had positive impacts on the empower-ment of women including delaying marriage and reducing fertility rates.Menon, Rodgers, and Kennedy (2013) and Newman, Tarp, and van denBroeck (2015) find positive impacts of land titling, and in particular jointland titling where women are included in the land registration, on welfareoutcomes for women and households more generally. Indeed, it is now widelyacknowledged that promoting gender equality within households and inparticular putting resources under the control of women, can significantlyimprove welfare and progress the development process (Duflo 2003). As such,in addition to gender equality being a human right, promoting gender equal-ity will also contribute to development through the impact that femaleempowerment has on the welfare of families, and children in particular, inrelation to, for example, nutrition, education, and agricultural productivity.2

In this chapter we use the Viet Nam Access to Resources Household Survey(VARHS) to analyse the extent of gender inequality in the welfare of house-holds and individuals living in rural areas.We consider the two distinct groupsof women living in rural Viet Nam. We first examine female-headed house-holds, the majority of which are headed by widows (68 per cent). Theseaccount for around 20 per cent of the VARHS sample and so represent asignificant proportion of rural households. Using the balanced panel of2,181 households, we examine the economic situation of female-headedhouseholds and consider how they are different to their male counterparts.We find that female-headed households are a very distinct socioeconomicgroup that is particularly vulnerable. Second, we focus our analysis on indi-viduals rather than households. We make use of the rich data collectedthrough VARHS on each individual within each household. We examine theeconomic status of women (adults) relative to men and examine how thewelfare of each group, relative to each other, has evolved over the 2008 to

2 See van den Bold, Quisumbing, and Gillespie (2013) for an overview of the evidence linkingfemale empowerment and child nutrition; and Doss (2013) for an overview of the literature linkingfemale empowerment to children’s education.

Gender Inequality and the Empowerment of Women

223

2014 period. We focus on three sets of outcomes, namely: health, education,and economic activities, and use a cohort analysis that allows us to comparethe characteristics of women and men within given age brackets over time.We conclude our analysis with an examination of the extent to which

female empowerment has taken place in Viet Nam and whether this has ledto increased household welfare outcomes. This analysis is motivated by theliterature mentioned earlier in this section which proposes that resources heldin the hands of women are good for economic development and in particularfor household and child welfare outcomes.Wemeasure female empowermentusing three measures: the proportion of income that a women earns fromwaged employment (on the assumption that this income is more likely to bekept by the woman), whether or not the woman is in charge of managing thehousehold land, and whether or not the woman has joint property rights tothe land that she and her spouse farm. Using the full panel dataset from 2008to 2014, and excluding female-headed households, we examine the relation-ship between these empowerment indicators and household consumption.The chapter is structured as follows. In Section 11.2 we examine the

characteristics of female-headed households in terms of socioeconomic char-acteristics, income, and vulnerability. In Section 11.3 we present a cohortanalysis using the individual level data focusingon four cohorts: 18–30-year-olds,31–45-year-olds, 46–60-year-olds, and those aged 61 and over. In Section11.4 we present measures of female empowerment and relate these measuresto household welfare. Section 11.5 concludes.

11.2 Characteristics of Female-Headed Households

Approximately one-fifth of households in the VARHS sample were headed bywomen. In this section we explore the characteristics of these households.Table 11.1 presents descriptive statistics for a variety of household character-istics disaggregated by the gender of the household head.Female-headed households were, on average, older than male-headed

households and were less likely to have children. They were also much lesslikely to be married and most female heads (68 per cent) were widows. Theywere also less likely to be ethnic minorities and were less likely to have tertiary-level education than male-headed households.3

3 In making other comparisons across these groups the distinct features of female-headedhouseholds should be borne in mind. Single male-headed households of a similar age would be anatural comparison group but there are not enough of these households in our sample to drawmeaningful conclusions. As such, we focus on comparing female-headed households with all othermale-headed households in the sample for the remainder of this section.


224

Table 11.2 presents descriptive statistics on the income and assets offemale-headed households compared to their male counterparts. Female-headed households were less well off than male-headed households. In allyears (monthly) income levels were significantly lower. While the incomelevels of female-headed households grew significantly between 2008 and2014, the gap between male- and female-headed households widened. In2014, the income of male-headed households was 27 per cent more thanfemale-headed households compared with a gap of 20 per cent in 2008.

Despite lower income levels female-headed households had similar levels offood expenditure per capita to male-headed households, and had even higherlevels in 2010. This could reflect the smaller average household size of female-headed households. It also suggests that where women have control overresources, general household welfare is higher, particularly relating to food

Table 11.1 Characteristics of female-headed households, 2008–14

2008 2010 2012 2014

Head of household Female Male Female Male Female Male Female Male

Age 44.12 39.20*** 46.40 40.68*** 47.74 41.95*** 50.30 44.15***Has children 0.41 0.52*** 0.45 0.56*** 0.40 0.51*** 0.39 0.49***HH size 3.75 4.78*** 3.47 4.57*** 3.40 4.47*** 3.36 4.39***Married 0.29 0.96*** 0.28 0.96*** 0.25 0.94*** 0.25 0.95***Has highereducation

0.10 0.18*** 0.12 0.21*** 0.10 0.21*** 0.13 0.23***

Ethnic minority 0.09 0.24*** 0.08 0.24*** 0.09 0.23*** 0.10 0.24***n 458 1,716 462 1,719 480 1,701 522 1,659

Note: *** indicates that the difference between female and male headed households is statistically significant at 1% level.

Source: Author’s calculations based on VARHS 2008–14 survey data.

Table 11.2 Household income and assets and female-headed households, 2008–14

2008 2010 2012 2014


Income (000 VND)a 4,949 5,949*** 5,823 7,058** 6,021 7,895*** 6,840 8,707***Food exp. p.c.(000 VND)

321 308 372 343** 462 444 463 452

Savings (000 VND) 20,213 21,256 30,693 31,952 32,910 43,678 35,941 39,487Loans (000 VND) 10,291 17,687 11,271 20,265*** 15,961 20,765 10,021 22,884**Durables (000 VND)b 3,577 4,524*** 3,878 5,186*** 4,005 5,693*** 3,849 5,616***Land area (ha) 4,500 8,837*** 4,244 8,615*** 4,636 8,509*** 4,302 8,288***Red book 0.85 0.86 0.85 0.80** 0.93 0.88*** 0.94 0.90***n 458 1,716 462 1,719 480 1,701 522 1,659

Note: a VND = Vietnamese Dong, 22,500 VND approximately equivalent to US$1; b Outliers in the top 99th percentile ofthe distribution of the value of durables in each year are removed; *** indicates difference between female and maleheaded households is statistically significant at 1% level and ** at 5% level.



225

and nutrition.4 This latter explanation could also account for the fact thatdespite differences in household income the savings levels of female-headedhouseholds were also similar to those of male-headed households. While theactual level was lower in each year the difference was not statistically signifi-cant at conventional levels.Female-headed households were worse off than their male counterparts in

terms of other assets. The value of their durable goods5 was much lower (sig-nificantly so in all years) and it appears that they had less access to credit, withmuch lower loan amounts than male-headed households. They also had muchsmaller land holdings (about half that of male-headed households). They were,however, more likely to have a red book (land-use certificate (LUC) for the landthat they own). This suggests that securing property rights is more importantfor female-headed households than male-headed households.Table 11.3 explores the income sources of female-headed households. They

were less likely to rely on agricultural income and (although to a lesser extent)income from waged employment than male-headed households. In 2008 and2010 they were more likely to earn income from household enterprises thanmale-headed households, but in 2012 and 2014 they were also less likely toearn income from this source. In terms of diversification, it is clear thatbetween 2008 and 2014 male-headed households became less specialized inagriculture and more diversified into other types of activities. There is noevidence that female-headed households exhibited a similar pattern. Thedecline in the participation of female-headed households in economic

Table 11.3 Sources of income and female-headed households, 2008–14

2008 2010 2012 2014


Agric. Income 0.82 0.91*** 0.79 0.88*** 0.75 0.86*** 0.73 0.85***HH enterprises income 0.64 0.57* 0.63 0.58*** 0.61 0.62*** 0.61 0.66**Wage income 0.25 0.29*** 0.19 0.30* 0.20 0.27 0.20 0.25*Agriculture only 0.19 0.27*** 0.22 0.23 0.19 0.21 0.18 0.20Diversified 0.74 0.72 0.72 0.75 0.71 0.76** 0.71 0.77***No activitiesa 0.06 0.01*** 0.06 0.02*** 0.10 0.03*** 0.11 0.03***N 458 1,716 462 1,719 480 1,701 522 1,659

Notes: a No activities refers to households that do not earn an income from any of the economic activities consideredhere. The main source of income for these households was from public and private transfers. *** indicates differencebetween female and male headed households is statistically significant at 1% level, ** at 5% level, and * at 10% level.


4 For evidence linking female empowerment to child nutrition see, for example, Fafchamps,Kebede, and Quisumbing (2009), Guha-Khasnobis and Hazarika (2006), Kennedy and Peters(1992), and Thomas (1990).

5 Durable goods include TVs, radios, computers, mobile phones, household appliances, motorvehicles, and farm assets.


226

activities over the sample period is likely due to the ageing of this groupbeyond the retirement age for women in Viet Nam (55 years), making itmore likely that they are not engaged in any economic activities. Indeed,when we focus on the sample of households that are under 55 in all periodsalmost all are engaged in some form of economic activity (result not shown).

In Table 11.4 the vulnerability of female-headed households to incomeshocks is compared to that of male-headed households. In all years female-headed households were less vulnerable to natural shocks than male-headedhouseholds. This is likely due to the fact that they have less land and are lesslikely to engage in agricultural activities, which are more affected by naturalshocks than other types of activities. There is some evidence, however, thatthey weremore vulnerable to economic shocks, particularly in 2008 and 2014.This reflects the underlying vulnerability of female-headed households giventhat the majority were widowed, surviving on much lower income levels thanother households.

It is clear from the analysis presented in this section that female-headedhouseholds in the VARHS sample were distinct from other households in anumber of different respects. They were low-income households typicallyheaded by widows. They had less land and were less engaged in agriculturalactivities than other households. They also had fewer assets more generally.They did, however, save as much as other households and had similar percapita food consumption levels, suggesting that they were equipped to copewith their lower standard of living. While the welfare of these householdsimproved between 2008 and 2014, this has not been to the same extent asother households. This makes them a vulnerable group, particularly in theface of unexpected income shocks.

11.3 Cohort Analysis

In this section we move away from focusing on female-headed households toexamine the situation of women more generally. VARHS gathers detailed

Table 11.4 Vulnerability of female-headed households, 2008–14

2008 2010 2012 2014


Natural shock 0.35 0.46*** 0.34 0.45*** 0.22 0.35*** 0.18 0.26***Economic shock 0.28 0.22*** 0.19 0.16 0.21 0.19 0.18 0.13***n 458 1,716 462 1,719 480 1,701 522 1,659

Note: *** indicates difference significant at 1% level, ** at 5% level, and * at 10% level.



227

information at the individual level for all household members. This allows usto explore how female household members compare to male householdmembers on a variety of different welfare measures and how their welfare, inabsolute and relative terms, has improved over time. We examine welfareoutcomes for four different cohorts: (i) 18–30-year-olds; (ii) 31–45-year-olds;(iii) 46–60-year-olds; and (iv) individuals over 60.We consider three broad measures of individual welfare. First, we consider

health outcomes using a general health indicator that records whether or notan individual suffered from any illness in the previous two weeks. For thoseindividuals who were ill we disaggregated by whether they suffered from achronic illness such as heart disease, respiratory disease or cancer, a mentalillness, or some temporary condition such as cold/flu or an injury. Second, weconsider two education outcomes: (i) whether the individual is literate; and(ii) the years of education attained by the individual. Third, we consider theeconomic activities of individual household members. We do not have infor-mation on the individual level of income of household members but we doknow the amount of time spent engaged in different types of economicactivities. We consider the number of days worked on aggregate and brokendown by type of activity, including days spent working in agriculture, collect-ing common property resources (for example, firewood and food grown oncommon property land), household enterprises, and waged employment. Thelatter two are more likely to be associated with an independent source ofincome for individuals and so we consider these superior from a welfareperspective.

11.3.1 Health Outcomes

Table 11.5 presents differences in health outcomes for men and women in theVARHS balanced panel for the 2008 to 2014 period. The incidence of illnessdeclined for both men and women between 2008 and 2014 across all cohorts.There is also a change in the type of illnesses reported, with both chronic andmental illnesses much more common in 2014 compared with 2008. Whilethis may be due to a higher incidence of these types of illnesses it could also bedue to better detection and reduced stigma. There are few statistically signifi-cant differences between males and females in the incidence of illness and thetypes of illnesses reported, particularly in 2014. In 2008, for example, males inthe 31–45, 46–60, and 60+ age groups were more likely to report that they hadbeen ill in the previous two weeks. In 2014 there was no gender difference.In terms of the type of illness, males in the 31–45 age group in 2014 weremuch less likely than females to report that they suffered from amental illness(26 per cent of ill men compared with 44 per cent of ill women).


228

Overall, it is clear that health outcomes improved for all between 2008 and2014 with no evidence of gender disparities.

11.3.2 Education Outcomes

Differences between 2008 and 2014 in education outcomes for male andfemale cohorts are presented in Table 11.6. In 2008, literacy rates were highfor both males and females among all but the oldest cohort. In all cases,women outperformed men with significantly higher rates. Between 2008and 2014 literacy rates did not change much in general. One exception wasa large improvement in literacy rates for males over 60 years old who startedout at a low rate of 63 per cent in 2008 climbing to 76 per cent in 2014.Females continued to outperform males on this measure in 2014 in all agecohorts.

There were significant increases in the years of schooling for both men andwomen in all age cohorts. The most notable improvements were among18–30-year-olds. Significant improvements for men are evident in the 46–60age group and in the over 60s. Again women outperformed men on thisoutcome across all age cohorts in both 2008 and 2014. One exception wasamong the 18–30 age group, where, in 2014, there was no statistical differencein the average years of schooling of men and women.

Table 11.5 Gender cohort analysis 2008–14, health outcomes

18–30 years 31–45 years

Female Male Female Male

Individual 2008 2014 2008 2014 2008 2014 2008 2014

Sick 0.06 0.03 0.06 0.03 0.09 0.05 0.13** 0.06Of which:Chronic illness 0.08 0.06 0.11 0.07 0.10 0.06 0.18 0.14Mental illness 0.16 0.28 0.08 0.27 0.20 0.44 0.17 0.26*Other illness 0.77 0.67 0.81 0.70 0.73 0.53 0.68 0.63n 1,121 1,102 987 947 923 731 1,009 740

46–60 years 61+ years

Sick 0.15 0.12 0.19* 0.11 0.26 0.25 0.32* 0.27Of which:Chronic illness 0.11 0.25 0.18 0.24 0.28 0.40 0.21 0.33Mental illness 0.18 0.21 0.17 0.23 0.29 0.25 0.22 0.22Other illness 0.72 0.59 0.68 0.65 0.46 0.46 0.64*** 0.54n 709 884 746 953 367 460 558 650

Note: *** indicates male and female outcomes statistically different at 1% level, ** at 5% level and * at 10% level.



229

Overall, there have been significant improvements in education across allage groups for bothmen and women. The former began from a lower base andsome of the gaps between men and women in educational outcomes wereclosed between 2008 and 2014, particularly for younger age cohorts.

11.3.3 Economic Activities

In the final part of the cohort analysis we examine differences in time useacross time and gender. We focus on the days worked in different types ofactivities, including agriculture, common property resources, householdenterprises, and waged work. Summary statistics are presented in Table 11.7.There were declines in the average number of days worked by men and

women in all cohorts. This is explained in large part by the decline in thenumber of days spent working on agricultural activities. At the same time theaverage number of days spent in waged employment increased for all cohortswhile the number of days spent in household enterprises increased for31–45-year-olds.Women worked significantly more days than men across all age cohorts.

The gap in the average number of days worked grew between 2008 and 2014for the 18–30 years cohort and the 46–60 years cohort. Women spent signifi-cantly more days in waged employment thanmen. In the 18–45 years cohortsthey also spentmore time collecting common property resources although theoverall number of days spent in this activity was low. Men, on the other hand,particularly those in the 31–45 years cohort, spent more days than womenengaged in agricultural activities.

Table 11.6 Gender cohort analysis 2008–14, education outcomes



Individual 2008 2014 2008 2014 2008 2014 2008 2014

Literate 0.96 0.98 0.93*** 0.94*** 0.91 0.90 0.87** 0.84***Years ofeducation

9.22 10.30 8.92** 10.11 7.12 7.85 6.43*** 6.96***

n 1,121 1,099 987 946 923 730 1,009 740


Literate 0.93 0.93 0.88*** 0.90** 0.89 0.92 0.63*** 0.76***Years ofeducation

7.22 7.94 5.87*** 7.01*** 5.60 6.77 2.41*** 4.12***

n 709 884 746 953 366 460 557 650

Note: *** indicates male and female outcomes statistically different at 1% level, ** at 5% level and * at 10% level.



230

It is not clear how the gender disparities in the economic activities of menand women might impact on welfare outcomes. On the one hand, the factthat women worked more days than men suggests that they face a greaterburden of responsibility for generating income thanmen. Given that the timeuse data do not consider the amount of time spent performing householdduties, the figures presented here could understate the gap between men andwomen. On the other hand, working for a wage could empower women byincreasing the resources under their control, potentially leading to betterwelfare outcomes for them and their families. Indeed, women in paid employ-ment were much more likely to work in service sector jobs than men in paidemployment. For the sample period as a whole, 55 per cent of women in paidemployment worked in the services sector compared with 34 per cent of men.Men were much more likely to take up employment in the agricultural sector(41 per cent) compared to women (27 per cent). We explore the extent towhich there is evidence that this empowered women in Section 11.4.

11.4 Female Empowerment and Welfare Outcomes

In this section we use the balanced panel of data to perform a household fixedeffects analysis of the impact of female empowerment on household welfareoutcomes measured in various ways. We consider three different measures of

Table 11.7 Gender cohort analysis 2008–14, economic activities



Individual 2008 2014 2008 2014 2008 2014 2008 2014

Total days worked 146 139 142 123*** 217 195 195*** 178***Days agric. 49 26 52 26 90 54 107*** 64***Days cpr 6 3 4** 3*** 8 6 6** 4**Days HH ent. 13 12 15 10 33 35 36 41Days wage 79 98 71** 86** 87 101 48*** 69***n 1,121 1,102 987 947 923 731 1,009 740


Total days worked 192 161 175*** 140*** 70 60 55** 49**Days agric. 101 62 112** 69** 47 31 39 26*Days cpr 6 5 4** 4** 2 3 2 2Days HH enterprises 31 27 39* 31 12 13 10 11Days wage 56 68 22*** 36*** 9 14 4** 10n 709 884 746 953 367 460 558 650

Note: *** indicates male and female outcomes statistically different at 1% level, ** at 5% level, and * at 10% level;cpr = common property resources.



231

female empowerment. First, following from the analysis presented inSection 11.3, we measure the extent of empowerment of female householdmembers as the proportion of total days worked by women that are in wagedemployment. Second, we use an indicator variable for whether a female in thehousehold is responsible for making decisions relating to the land that isowned by the household. Third, we use an indicator variable for whether afemale’s name is listed in the household red book. We restrict our analysis tohouseholds that are not headed by a female to ensure that we are capturingintra-household effects of female empowerment.Table 11.8 presents summary statistics for the evolution of these variables

among the (balanced) VARHS sample of male-headed households over thefour years. Increases in female empowerment measures are evident on mostindicators. In particular, consistent with the story presented in Section 11.3,we find waged work made up a greater proportion of women’s income in eachyear. Between 2008 and 2010 the number of households where a femalehousehold member made decisions in relation to the management of theland increased from 37 per cent to 41 per cent. There has, however, been noincrease in this measure since 2010. The proportion of households where awoman was named on the LUC increased significantly between 2008 and2012 from around 11 to 17 per cent. By 2014, however, this proportion haddeclined to 2010 levels. Overall, these summary indicators provide someevidence of an improvement in female empowerment since 2008 but muchless so in later years of the sample.In the final part of our analysis we explore the impact of female empower-

ment on household welfare. We use household expenditure on food as anindicator of welfare in our analysis. Food expenditure is generally considered amore reliable and accurate measure of welfare than household income giventhat it is less likely to be under-reported and is less likely to suffer frommeasurement error. The variable is constructed by aggregating the value of aset of food items consumed by the household in the previous month and isconverted to real terms using a national food price index. To explore therelationship between female empowerment and household welfare on thismeasure, we estimate the following econometric model:

Table 11.8 Indicators of female empowerment, 2008–14

Empowerment indicator 2008 2010 2012 2014

Proportion wage work women 32.17 34.38 36.22 39.24Female manager 37.06 41.01 40.75 40.66Joint property rights 10.98 11.52 17.14 11.91n = 1,584 households in each year



232

welht ¼ βXht þ δ1empowerht þ αh þ τt þ εiht

where welht is the welfare measure (food consumption per capita) for house-hold h in time t; Xh is a vector of household-specific variables, includingcharacteristics of the household head, income, land ownership, the presenceof a household enterprise, and the incidence of natural and economic incomeshocks; empower represents the three different measures of female empower-ment; αh are household fixed effects that absorb all time-invariant household-specific characteristics such as, for example, the ethnicity of the householdhead; τt are time dummies; and εiht is a statistical noise term.

The results are presented in Table 11.9. Column (1) describes the relation-ship between various household characteristics and food expenditure beforeany of the empowerment indicators are included. Most of the results forthese control variables are as expected. Household consumption per capitawas lower in bigger households and higher in households with moreincome. Assets were also highly correlated with household consumption:both durable goods and having an LUC or ‘red book’ were positively asso-ciated with food consumption per capita. One, perhaps surprising, result isthat households that experienced economic shocks actually consumed moreper capita than other households. It could be that richer households bothconsume more and have more assets that are vulnerable to shocks. Alterna-tively, it suggests that the coping strategies of these households in the faceof economic shocks are more than adequate to ensure consumptionsmoothing. It should be noted that the sample considered here excludesfemale-headed households, which, as seen in Section 11.2, are a particularlyvulnerable group.

In column (2) we add the first empowerment indicator, namely the propor-tion of total days worked by women in waged employment. We find a positiveand well-determined relationship, which suggests that the greater the propor-tion of a woman’s time spent working for a wage, the greater the household’slevel of per capita food expenditure. This is unlikely to be an income effect as itis controlled for in the regression. The magnitude of the effect is economicallymeaningfully. A 1 percentage point increase in the proportion of days workedby women in waged employment leads to an 8 per cent increase in per capitahousehold consumption. This is about one third of the magnitude of theimpact of a 1 per cent increase in household income on per capita householdconsumption.

In column (3), the second welfare measure is considered, namely whether ornot a woman in the household manages the land. A similar result emerges.In column (4) we find a similar effect of a woman in the household beingincluded in the land title or red book. In column (5) we include all measuressimultaneously and find that all three results hold, suggesting that each


233

empowerment measure has its own independent effect on household welfare.It should be noted that each model controls for differences in income, assets,marital status, age, presence of children, exogenous shocks, general trendsin household welfare, and all time-invariant household characteristics. Evenwhen these factors are controlled for, households where women areempowered have a higher level of welfare. While caution should be exercisedin interpreting these results as causal, these findings provide some evidencethat female empowerment and household welfare go hand in hand.

Table 11.9 Female empowerment and welfare, food consumption per capita

(1) (2) (3) (4) (5)

Empowerment measuresProportion wage work women 0.089*** 0.083***

(0.031) (0.031)Female manager 0.042** 0.048**

(0.019) (0.019)Joint property rights 0.049** 0.047**

(0.023) (0.023)Household characteristicsAge 0.012 0.013 0.011 0.012 0.013

(0.008) (0.009) (0.008) (0.008) (0.009)Age2 �0.000 �0.000 �0.000 �0.000 �0.000

(0.000) (0.000) (0.000) (0.000) (0.000)Married 0.007 �0.001 0.004 0.007 �0.004

(0.068) (0.083) (0.068) (0.068) (0.083)Children 0.019 0.028 0.020 0.019 0.029

(0.028) (0.029) (0.029) (0.028) (0.029)Higher education 0.015 0.016 0.014 0.014 0.014

(0.034) (0.037) (0.034) (0.034) (0.036)HH size �0.068*** �0.075*** �0.068*** �0.068*** �0.075***

(0.012) (0.012) (0.012) (0.012) (0.012)Income (log) 0.239*** 0.225*** 0.239*** 0.238*** 0.224***

(0.016) (0.015) (0.016) (0.016) (0.015)Loans (log) 0.002 0.002 0.002 0.002 0.002

(0.002) (0.002) (0.002) (0.002) (0.002)Land area (log) 0.026 0.010 0.025 0.026 0.010

(0.020) (0.021) (0.020) (0.020) (0.021)Household enterprise 0.028 0.053** 0.026 0.028 0.051**

(0.022) (0.024) (0.023) (0.022) (0.024)Durables (log) 0.037*** 0.038*** 0.037*** 0.037*** 0.037***

(0.010) (0.011) (0.010) (0.010) (0.011)Red book 0.100*** 0.098*** 0.099*** 0.092*** 0.088***

(0.032) (0.033) (0.032) (0.032) (0.033)Natural shock 0.008 0.010 0.008 0.008 0.010

(0.018) (0.018) (0.018) (0.018) (0.018)Economic shock 0.057*** 0.058*** 0.056*** 0.058*** 0.057***

(0.020) (0.021) (0.020) (0.020) (0.022)Observations 6,574 6,174 6,574 6,574 6,174Number of HH 1,774 1,715 1,774 1,774 1,715

Notes: Each model includes household and time fixed effects. Robust standard errors clustered at the household level inparentheses. *** indicates statistical significance at 1% level, ** at 5% level and * at 10% level.



234

11.5 Conclusion

Viet Nam has made significant progress in relation to gender equality. How-ever, as this chapter reveals, significant gaps remain. Using data from theVARHS for 2008, 2010, 2012, and 2014 we examine gender differences inthe welfare of Vietnamese households and individuals and how they haveevolved over this period.

Our analysis reveals that female-headed households are a distinct groupwithin VARHS with very different characteristics from other households.They are low-income households and a large proportion of them are headedby widows. They have less land and are less engaged in agricultural activitiesthan other households. Their welfare has improved over the period of analysisbut not to the same extent as other households. In particular, they are morevulnerable to income shocks than male-headed households.

Focusing on the panel of individuals within VARHS households we per-formed a cohort analysis examining differences in the welfare of women andmenwithin specified age groups andhowthese changedover time.Anumber ofinteresting findings emerge. First, we find that education outcomes improvedfor both men and women. In general, women outperformed men on literacyand years of education but this gap is closing over time. Second, we foundoverall declines in the number of days spent working in agricultural activitiesand an increase in days spent in waged employment for bothmen andwomen.This is consistentwith theongoing structural transformation in theVietnameseeconomy. Interesting from a gender perspective, however, is that women spentmore days working than men in all age cohorts, mainly due to significantlymore days spent in waged employment, Moreover, for 18–30-year-olds and46–60-year-olds this gap has widened over the sample period.

The last part of our analysis focused on indicators of female empowermentand the extent to which there is evidence of: (i) an increase in femaleempowerment over the 2008 to 2014 period; and (ii) whether femaleempowerment is associated with higher levels of household welfare as meas-ured by food expenditure per capita. We find on the basis of three empower-ment indicators (proportion of time spent in waged employment, whetherwomen are involved in land management decisions within the household,and whether land is jointly titled in a female household member’s name) that,in general, women were more empowered in 2014 than in 2008 but that theempowerment indicators have remained relatively static in the last few years.We find, though, a strong correlation between each indicator and householdfood expenditure per capita, suggesting an important link between empower-ing women and household welfare.

Overall, our findings suggest that efforts to promote gender equality,through, for example, the law on gender equality, should be stepped up to


235

avoid stagnation in the progress already made. Moreover, building capacityfor the empowerment of women by providing women with more agency aswell as more resources has the potential to progress economic development ina significant way.

References

Doss, C. (2013). ‘Intra-Household Bargaining and Resource Allocation in DevelopingCountries’. World Bank Research Observer, 28(1): 52–78.

Duflo, E. (2003). ‘Grandmothers and Granddaughters: Old Age Pension and Intra-Household Allocation in South Africa’. World Bank Economic Review, 17(1): 1–25.

Fafchamps, M., B. Kebede, and A. Quisumbing (2009). ‘IntrahouseholdWelfare in RuralEthiopia’. Oxford Bulletin of Economics and Statistics, 71(4): 567–99.

Guha-Khasnobis, B. and G. Hazarika (2006).Women’s Status and Children’s Food Securityin Pakistan. WIDER Working Paper 2006/03. Helsinki: UNU-WIDER.

Heath, R. and A.M.Mobarak (2015). ‘ManufacturingGrowth and the Lives of BangladeshiWomen’. Journal of Development Economics, 115: 1–15.

Jensen, R. (2012). ‘Do Labor Market Opportunities Affect Young Women’s Work andFamily Decisions? Experimental Evidence from India’. Quarterly Journal of Economics,127(2): 753–92.

Kennedy, E. and P. Peters (1992). ‘Household Food Security and Child Nutrition: TheInteraction of Income and Gender of Household Head’. World Development, 20(8):1077–85.

Menon, N., Y. Rodgers, and A. Kennedy (2013). ‘Land Rights and Economic Security forWomen in Viet Nam’. World Bank Working Paper. Washington DC: World Bank.

Newman, C., F. Tarp, and K. van den Broeck (2015). ‘Property Rights and Productivity:The Case of Joint Titling in Viet Nam’. Land Economics, 91(1): 91–105.

Thomas, D. (1990). ‘Intra-Household Resource Allocation: An Inferential Approach’.Journal of Human Resources, 25(4): 635–64.

Van den Bold, M., A. Quisumbing, and S. Gillespie (2013). Women’s Empowerment andNutrition: An Evidence Review. IFPRI Discussion Paper 01294. Washington DC: IFPRI.

World Bank (2011). Viet Nam Country Gender Assessment. Washington DC: World Bank.World Bank (2016). Vietnam 2035: Toward Prosperity, Creativity, Equity and Accountabil-ity. Washington DC: World Bank.


236

12

Children and Youths

Gaia Narciso and Carol Newman

12.1 Introduction

As demonstrated throughout this book, the ongoing process of structuraltransformation in rural Viet Nam has led to rising incomes and a diversifica-tion of economic activities away from agriculture. As incomes rise and ruralhouseholds become better off, the welfare of children, like other householdmembers, are likely to improve. Improvements in the level and security ofhousehold income (see Chapter 10) are likely to translate to improvements inthe health, educational attainment, and life opportunities of children andyoung people more generally. Moreover, as households shift out of agriculturetowards waged employment (see Chapter 5), children are less likely to spendtheir time on agricultural work allowing more time for school and study.Where economic transformation empowers women, this is also likely toimpact positively on the welfare of children (see Chapter 11).1 It is also pos-sible, however, that if the process of structural transformation has left somegroups behind or there are inequalities in the distribution of the fruits ofeconomic growth (see Chapter 10), children and youths, as a particularlyvulnerable group, are likely to be adversely affected.

In this chapter, we use the Viet Nam Access to Resources Household Survey(VARHS) data to examine how the lives of the children and youths living inrural Viet Nam have been impacted by structural transformation. First, weexamine the characteristics of households with children compared to those

1 A large literature exists, which highlights how resources in the hands of women are more likelyto be used to improve children’s outcomes, particularly girls, than resources held in the hands ofmen (Pitt and Khandker 1998; Duflo 2003; Qian 2008). Newman (2015) shows significantimprovements in the empowerment of women in Viet Nam over the last decade, evidentthrough an expansion in access to resources and economic opportunities for women.

without and how these have changed between 2008 and 2014. Second, weexploit the detailed individual level data contained in VARHS on a range ofdifferent welfare measures to compare different age cohorts over time toexamine whether children in general are doing better in 2014 comparedwith 2008. We measure the welfare of children using information on theirhealth, education attendance and attainment, as well as engagement in labour(agricultural, household enterprise, and waged employment). We also exam-ine whether there is evidence in our data of heterogeneity in welfare gainsalong both gender and ethnic lines. The latter is of particular relevance giventhat the World Bank’s (2016) report Vietnam 2035 highlights the significantgaps in economic opportunities for children in poor households, and inparticular among ethnic minorities.In the third part of our analysis, we create a panel dataset of households that

contains individual level information on children that tracks each child presentin each household in 2008 through each round of the survey up to 2014. Thisallows us to determine the dominating household characteristics in determin-ing thewelfare of children over the period. Finally, we examine whether there isevidence that female empowerment and an increase in the resources held in thehands of women is linked with improvements for children.Early studies have analysed the relationship between economic develop-

ment and child wellbeing in Viet Nam, in particular with respect to childlabour. Using data from the Viet Nam Living Standards Surveys (VLSS) for theperiod 1993 to 1998, Edmonds (2005) shows a significant drop in child labourof about 30 per cent over a five-year period. Given the panel nature of the dataused, the author is also able to disentangle the different determinants of thereduction in child labour. The author finds that improvements in householdeconomic status explain a stark 60 per cent of the change in child labour overthe period considered. In particular, the effect of improvements in economicstatus on reducing child labour is found to be greater in poorer householdsthan in wealthier ones. These results support the findings of the cross-countryliterature that suggests a strong relationship between gross domestic product(GDP) per capita and child labour (Krueger 1997).Edmonds and Turk (2004) further explore heterogeneity in the incidence and

drop in child labour inVietNamusing theVLSS also for the period1993 to1998.In particular, girls experience a smaller decline in child labour than boys.Children living in rural areas are also found to be more likely to work thanchildren in urban areas, in particular in traditional occupations. Parents’ busi-ness activities are linked to child labour, as child labour ismore likely to increasewith the opening and closing of household enterprises. Finally, children ofethnic minorities are found to be more likely to work than children of non-ethnic minorities. Overall, Edmonds and Turk provide evidence of a strongassociation between poverty and child labour and highlight the importance of


238

anti-poverty programmes as a path to reducing child labour. Edmonds andPavcnik (2005a) investigate the impactof the integrationofVietNam’s ricemarketon child labour and provide evidence that the increase in rice prices between1992 and 1993 and 1997 and 1998was linked to a decrease in child labour.

Beegle, Dehejia, and Gatti (2009) use VLSS to analyse the effects of childlabour on education, wages, and health. They provide evidence that childlabour has a negative effect on school participation and education attainmentfive years after the child’s labour experience. Young adults involved in labourduring their childhood are found to have higher wages. However, this effect isreversed over a longer time period, as the earnings’ loss due to lower educationattainment exceeds the initial wage gain due to child labour. While Beegle,Dehejia, andGatti (2009) find no impact of child labour on health, O’Donnell,Rosati, and van Doorslaer (2005) find a negative impact of child labour ongirls’ health, five years after the child labour episode.

We contribute to the existing literature by providing evidence of the pro-gress made in Viet Nam towards improved child wellbeing in recent years.Section 12.2 presents descriptive statistics on the characteristics of householdswith children in the VARHS sample. In Section 12.3, a cohort analysis isconducted to determine how the welfare of children of the same age in 2008compares to that of children in 2014. The cohort analysis is also disaggregatedalong gender and ethnic lines. Panel data analysis linking household charac-teristics and female empowerment to children’s welfare outcomes is presentedin Section 12.4. Section 12.5 concludes.

12.2 The Characteristics of Households with Children

In 2014, 54 per cent of households in the VARHS sample had children.2 Of thehouseholds with children, the average number of children was 1.68 (0.81 girlsand 0.87 boys). Fertility rates in general appear to have increased over thesample period. In 2008, 49 per cent of the VARHS sample had children, withthese households having an average of 1.67 children (0.82 girls and 0.85boys). It should be noted, however, that these statistics are based on theunbalanced panel of households which includes the addition of over 500new younger households in 2012 to account for ageing of the originalVARHS households sampled in 2006. The small increase in the proportion ofhouseholds with children is likely accounted for by these households.

Table 12.1 explores the variation in fertility across seven different regionscovered by VARHS, namely: Red River Delta (Ha Tay), North (Lao Cai, Phu

2 Any household member under the age of 18. We consider different age brackets throughoutthe analysis: 6–9-year-olds, 10–15-year-olds, and 15–18-year-olds.

Children and Youths

239

Tho, Lai Chau, and Dien Bien), Central Coast (Nghe An, Quang Nam, andKhanh Hoa), Central Highlands (Dak Lak, Dak Nong, and Lam Dong), andMekong River Delta (MRD) (Long An).3 The table presents the proportion ofhouseholds in the VARHS sample in each region that have children and forthose households the average number of children.The proportion of households with children is highest in the Central

Highlands and in the North. While the proportion of households with chil-dren increasedmarginally in the other regions between 2008 and 2014, in partdue to the addition of new younger households to the sample in 2012, thedifference between the Central Highlands, the North, and the rest of thecountry is still quite large in 2014. Moreover, households with children inthe Central Highlands and the North have more children on average thanhouseholds with children in other regions. For example, in 2008, these house-holds had on average 1.81 children compared with an average of 1.53 childrenfor households with children in other regions. The gap closes a little between2008 and 2014 at an average of 1.78 and 1.59, respectively, in 2014.We explore the characteristics of households with children in Table 12.2. In

each year we test for the statistical significance of the difference in the averagevalue of each variable for households with children and households without.The head of household in households with children is on average younger

than in households with no children and is also more likely to be married.They are also less likely to be headed by a woman. In 2010, heads of house-holds with children were significantly less likely to have higher education(i.e. post-secondary schooling) than in households with no children.With theaddition of new younger households to the sample in 2012, this differencedisappears. Ethnic minority households are more likely to have children than

Table 12.1 Geographical variation in fertility

2008 2010 2012 2014

% HHwithchildren

Meanno. ofchildren

% HHwithchildren

Meanno. ofchildren

% HHwithchildren

Meanno. ofchildren

% HHwithchildren

Meanno. ofchildren

Red River Delta 0.44 1.62 0.46 1.63 0.54 1.66 0.52 1.72North 0.52 1.71 0.57 1.72 0.59 1.77 0.57 1.73Central Coast 0.47 1.58 0.50 1.62 0.52 1.65 0.50 1.64Central Highlands 0.63 1.92 0.65 1.98 0.65 1.85 0.61 1.83Mekong River Delta 0.44 1.38 0.50 1.45 0.49 1.43 0.50 1.41

Note: unbalanced panel of households.

Source: Authors’ calculations based on survey data from VARHS 2008–14.

3 It should be noted that our data are not representative of the regions but the rural provinceswithin each region.


240

Table 12.2 Characteristics of households with children, 2008–14

2008 2010 2012 2014

HH withchildren

HH nochildren

HH withchildren

HH nochildren

HH withchildren

HH nochildren

HH withchildren

HH nochildren

Agea 37.45 43.49*** 38.55 45.96*** 37.34 46.65*** 38.36 48.25***Marrieda 0.85 0.77*** 0.84 0.78*** 0.85 0.75*** 0.85 0.74***Femalea 0.18 0.26*** 0.18 0.25*** 0.17 0.26*** 0.19 0.28***Higher ed.a 0.15 0.17 0.17 0.22*** 0.22 0.21 0.25 0.24Ethnic min.a 0.28 0.14*** 0.27 0.13*** 0.26 0.14*** 0.25 0.15***Income (000 VND) 1,111 1,600*** 1,400 2,054*** 1,665 2,267*** 1,771 2,423***Food exp. p.c. (000 VND) 260 359*** 305 403*** 396 522*** 402 516***Savings (000 VND) 21,327 21,057 31,536 31,505 36,539 43,164 37,061 38,162Loans (000 VND) 15,114 17,476 20,613 15,767* 26,044 15,617** 24,822 14,218**Durables (000 VND)b 4,469 4,127 5,240 4,491*** 5,389 4,937* 5,618 4,666***Land area (ha) 8,590 7,016*** 8,558 6,592*** 7,110 6,733 7,084 6,351Red book 0.83 0.89*** 0.77 0.86*** 0.78 0.89*** 0.84 0.92***Ag. income 0.90 0.86*** 0.87 0.84** 0.84 0.79*** 0.83 0.78***Wage income 0.60 0.56* 0.63 0.54*** 0.70 0.57*** 0.75 0.59***HHEnt. income 0.28 0.26 0.30 0.26** 0.30 0.22*** 0.29 0.21***Agriculture only 0.26 0.25 0.20 0.26*** 0.15 0.23*** 0.13 0.23***Diversified 0.74 0.70* 0.79 0.68*** 0.84 0.68*** 0.86 0.68***Nat. shock 0.45 0.41** 0.47 0.37*** 0.32 0.26*** 0.24 0.22Econ, shock 0.25 0.21** 0.16 0.17 0.18 0.19 0.12 0.15*n 1,125 1,161 1,195 1,050 1,532 1,228 1,471 1,254

Notes: unbalanced panel of households; a refers to household head; b Outliers in the top 99th percentile of the distribution of the value of durables in each year are removed;*** indicates difference significant at 1% level, ** at 5% level, and * at 10% level.


Kinh households. This is not surprising given the geographic differencespresented in Table 12.1, which show the highest fertility in the north-west,where over 87 per cent of households in the sample are ethnic minorities.The average monthly per capita income of households with children is lower

than in households without children. This is also reflected in the fact thathouseholds with children have lower food expenditure per capita in all years.4

In relation to assets, there is no statistically significant difference in thesavings levels of households with children compared to those without in anyyear, but in 2010, 2012, and 2014 households with children have significantlymore durable goods. They also own more land than households withoutchildren, at least in 2008 and 2010, but are significantly less likely to hold aland-use certificate (LUC) or red book for that land. On the whole it does notappear that households with children are wealthier than households without.They do however have more access to credit with a higher level of loans thanhouseholds without children in the later years of the sample.In terms of sources of income, households with children are significantly

more diversified and are more likely to earn income from all sources; agricul-ture, wage, and household enterprises. This may be due to the availability oflabour resources that allow them to engage in many different activities or maybe a means of managing risk. Indeed, households with children are morevulnerable to natural shocks which primarily affect agricultural productionbut are less likely to suffer from economic shocks associated with unemploy-ment or illness, suggesting that there are risk-coping mechanisms at work.

12.3 Cohort Analysis

The VARHS collects detailed information on all individuals in each householdincluding certain information on children. Using these data we can examinehow children’s welfare has evolved over the 2008–14 period. We consider threedifferent age cohorts in our analysis: 6–9-year-olds; 10–14-year-olds; and15–18-year-olds.5 We compare the welfare of children in each cohort in 2008with their counterparts in 2014. To ensure our sample is as close as possible tobeing representative we use the unbalanced panel of data so that the data in2014 capture the new younger households that were added in 2012.

4 Food expenditure items include pork, beef, chicken, fish, shrimp, fruit, sweets/biscuits,powdered or canned milk, liquid milk, beer, rice wine or other alcoholic drink, coffee, industrialbeverages, processed foods, and eating and drinking outside the home.

5 We do not report the characteristics of 0–5-year-olds, as we do not find any significant changeover time. Six years old is a natural starting age, given that this is the age that most children startschool in Viet Nam.


242

We consider three broad categories of child welfare: health, education, andchild labour. First, in relation to health, for each individual in the household,the survey respondent is asked whether that individual was ill in the previoustwo weeks. For those that were ill, they are then asked whether they were ill asa result of a range of illnesses, which we aggregate into chronic illness (includ-ing heart disease, respiratory illness, and cancer), mental illness, or other typesof temporary illness including colds, flu, other injuries, and so on. Second,in relation to education, we consider an indicator for whether children attendschool and for those above 4 years of age, how many years of educationthey have attained. Third, in relation to child labour, VARHS records detailedtime use data for all household members. The head of household records howmany days in the last year each household member worked in different typesof activities. They include agriculture, common property resources, workingfor the household enterprise, and working for a wage outside of the home.

Basu, Das, and Dutta (2010) and Edmonds and Pavcnik (2005b) highlightthe importance of including domestic work as child labour. Unfortunately,the VARHS data do not measure domestic work and household chores in aconsistent way over time, and so we cannot include them in our analysis. Weare aware that, by excluding domestic work from our analysis, we may under-estimate girls’ involvement in labour.

Table 12.3 presents each of these welfare measures for the three cohorts ofchildren in 2008 and 2014. The proportion of children in each cohort that arefemale is also presented. Girls represent about half the sample in each periodsuggesting that, at least in our sample, there is no evidence of an imbalance insex ratios.

Table 12.3 Characteristics of different cohorts of children, 2008–14

Cohort: 6–9-year-olds 10–14-year-olds 15–18-year-olds

Year 2008 2014 2008 2014 2008 2014

Female 0.50 0.53 0.50 0.51 0.54 0.50Sick in last 2 weeks 0.09 0.06* 0.07 0.03*** 0.07 0.03***Of which:Chronic illness 0.12 0.30** 0.09 0.04 0.16 0.10Mental illness 0.02 0.05 0.09 0.00 0.09 0.10Other illness 0.88 0.65*** 0.85 1.00** 0.75 0.81Attends school 0.69 0.76*** 0.91 0.97*** 0.64 0.75***Years of education 2.07 2.16 5.77 5.91* 8.93 9.58***Total days of work 6.26 1.85*** 21.34 6.70*** 64.64 34.40***Days’ work ag. 4.28 1.28*** 17.23 5.16*** 38.55 15.23***Days’ work cpr 0.52 0.16** 1.63 0.53*** 3.81 1.92***Days’ work ent. 0.00 0.00 1.04 0.49* 4.41 2.04**Days’ work wage 1.47 0.41 1.46 0.53 18.14 15.21n 561 606 1,028 836 1,071 738

Notes: unbalanced panel of households. *** indicates difference significant at 1% level, ** at 5% level, and * at 10% level.


Children and Youths

243

There was a decline in the proportion of children that experienced an illnessin the previous two weeks. This is somewhat suggestive of an improvement inthe health of children and young people over time. In the 6–9 age group therehas been a statistically significant increase in the diagnosis of chronic illnesses.The World Bank (2016) also reports a high incidence of respiratory infectionsamong children in Viet Nam due to the deterioration in environmental quality.The increase in chronic illness is not observed in other cohorts.Children are significantly more likely to attend school in 2014 compared

with 2008 in all age cohorts, and children over 10 also have, on average, moreyears of schooling. There have also been some improvements for childrenin terms of time use. Children spend considerably fewer days working at allactivities in 2014 compared with 2008 in all age groups. Of particular note isthe decline in the number of days children spend doing agricultural activitiesfrom 4.3 to 1 day a year in the 6–9-year-old age group, from 17.2 to 5.2 in the10–14-year-old age group, and from 38.5 to 15.2 in the 15–18-year-old agegroup. The number of days worked in waged employment are low for childrenaged under 15, and have declined over time to on average half a day a year.Declines in waged work are also evident among the older age group from18.1 days a year in 2008 to 15.2 in 2014, but the difference is not statisticallysignificant.Within this age groupmost jobs are in the services sector (43 per centin 2014), followed by manufacturing (32 per cent in 2014), and agriculture andother primary sectors (23 per cent in 2014).Overall, these statistics suggest that the welfare of children, in the areas of

health, education, and child labour, has improved between 2008 and 2014.These results seem to support the findings of the literature presented in theintroduction showing a positive trajectory of child wellbeing in Viet Namover time.Are these improvements homogenous across expenditure quintile? The

nature of the VARHS data allows children of the same age cohort to befollowed over time. We focus on 6–9-year-olds at the time of the 2008 surveyandwe investigate their school attendance and educational attainment for thefollowing three rounds of the survey. We divide the 6–9-year-olds cohort byexpenditure quintile, as measured in 2008. Table 12.4 reports the results.In 2008, only 51 per cent of the children in the bottom quintile attendedschool, versus 65 per cent of the children in the top quintile. Children in thetop quintile had already accumulated almost one more year of schoolingcompared to children in the bottom quintile. While school attendanceincreases for all groups over time, the difference between the top and bottomquintile remains quite large; only 57 per cent of the children in thebottom quintile attend school in 2014, while 76 per cent of the children inthe top quintile are in school. Interestingly, themiddle quintiles seem to catchup over time. In particular, the middle and second richest quintiles show a


244

significant increase in the level of school attendance, with around 74 per centof children attending school in 2014 compared with only 66 per cent in 2008.Moreover, while all groups improved their outcomes over time, the bottomquintile, that is, the children belonging to the poorest segment of society in2008, do not catch up with the other groups. A divergence in human capitalaccumulation between the poorest group and the rest may in fact prolongwelfare differences over time, making it more difficult for them to catch up inthe long run.

In the next step of our analysis we disaggregate cohorts into girls and boys.In the light of the findings by Edmonds and Turk (2004) on the heterogeneityin child wellbeing, we try to determine whether there are any gender dispar-ities in the distribution of welfare gains. We focus on the 6–18-year-old agegroups. The disaggregation is presented in Table 12.5 for the overall healthindicator, the education measures, and time use.

The incidence of sickness declined for both boys and girls between 2008 and2014. Boys are less likely than girls to have experienced an illness in theprevious two weeks. This difference, however, is only statistically significantin the 6–9-year-old age group. The school attendance rate and years of educa-tion completed increased or stayed the same between 2008 and 2014 for bothboys and girls across every cohort. In 2014, girls aged 6–9 years are signifi-cantly more likely to attend school than boys. Boys, however, have more yearsof education than girls in this age cohort. Boys also had significantly moreyears of education than girls in the 10–14-year-old age group in 2014. Thesefindings suggest that, while both girls and boys have experienced improve-ments in health and schooling outcomes, these gains have been particularlybeneficial for boys.

The decline in the number of days children spend working is also evidentacross both girls and boys, but with boys experiencing bigger gains in the olderage groups. In 2014, boys aged 15–18 years spend significantly fewer days than

Table 12.4 Evolution of education outcomes for children aged 6–9 in 2008, by foodexpenditure quintile in 2008

2008 2010 2012 2014

Quintile2008

Attendschool

Years ofeducation

Attendschool

Years ofeducation

Attendschool

Years ofeducation

Attendschool

Years ofeducation

1 0.51 3.96 0.61 4.45 0.58 5.40 0.57 6.592 0.63 4.59 0.70 5.67 0.69 6.40 0.67 7.503 0.66 4.67 0.76 5.57 0.77 7.05 0.74 8.104 0.66 4.74 0.72 5.86 0.79 7.05 0.73 7.995 0.65 4.93 0.76 5.68 0.79 6.81 0.76 7.90


Children and Youths

245

girls of the same age working outside of the household. This trend is alsoreflected in the number of days worked in different types of activities. Girls inthe 6–9-year-old age group spend more days working in agriculture than boysin 2008, but both experienced a decline in the average number of days toaround one per annum in 2014. Similarly, in the 10–14 age group, girlsexperienced a decline in the number of days worked in agriculture from 14.8to 5.13 between 2014 and 2008 compared with a decline from 19.6 to 5.2 forboys of the same age. In the 15–18 age group, the relative gains are evengreater for boys, with a decline in the average number of days worked inagriculture from 37.1 to 13.1 compared with a decline for girls of the sameage from 39.8 to 17.4. Boys spend, on average, more days working for a wagethan girls across all age groups, but this difference is not statistically signifi-cant. Overall, these trends suggest that while the welfare of both girls and boysimproved between 2008 and 2014, boys benefited to a greater extent thangirls. These findings are in line with the cross-country evidence, according towhich the prevalence of child labour is greater for girls and for boys (Edmondsand Pavcnik 2005a).Given the existing evidence of heterogeneity in child wellbeing with

respect to ethnicity (Edmonds and Turk 2004; World Bank 2016), we alsodisaggregate the cohort analysis by ethnicity of the household head.6 Descrip-tive statistics are presented in Table 12.6. There is no evidence of differences inhealth outcomes for children in ethnic minority households. Educationaloutcomes of ethnic minority households are also similar to those of Kinhhouseholds in younger age groups. Gaps, however, begin to emerge in olderage groups. The participation rate of children in ethnic minority householdsin education is significantly lower among 10–14-year-olds and 15–18-year-olds. In the case of the latter the difference is particularly stark, with only59 per cent of children from ethnic minority households attending schoolcompared with 81 per cent for Kinh households. Similarly, the average yearsof schooling attained by children over the age of 10 are significantly lower inethnic minority households. In 2014, the average years of schooling attainedby children in ethnic minority households in the 10–14-year-old age groupwas 5.6 compared to six for children in Kinh households. In the 15–18-year-old age group, children of ethnic minority households have an average of8.4 years of schooling compared to 10.1 for children in Kinh households.While living standards have increased over time for both Kinh and non-Kinh groups, it appears that a substantial difference in the level of welfarestill remains between the two groups. The lower human capital accumulation

6 A full analysis of ethnic minority households is provided in Chapter 13.


246

Table 12.6 Characteristics of different cohorts by ethnicity of household head, 2008–14

6–9-year-olds 10–14-year-olds 15–18-year-olds

2008 2014 2008 2014 2008 2014

Ethnicminority

Kinh Ethnicminority

Kinh Ethnicminority

Kinh Ethnicminority

Kinh Ethnicminority

Kinh Ethnicminority

Kinh

Sick in last 2 weeks 0.08 0.09 0.056 0.06 0.05 0.07 0.04 0.03 0.08 0.07 0.04 0.02Attends school 0.68 0.69 0.74 0.76 0.81 0.96*** 0.94 0.99*** 0.45 0.72*** 0.59 0.81***Years of education 1.82 2.20*** 2.18 2.16 5.05 6.09*** 5.62 6.03*** 7.46 9.53*** 8.36 10.12***Total days of work 9.24 4.70 2.02 1.78 30.60 17.14*** 13.56 3.85*** 82.47 57.44*** 42.69 30.73**Days’ work ag. 7.82 2.42*** 1.77 1.09 26.81 12.87*** 11.61 2.46*** 64.01 28.27*** 30.05 8.69***Days’ work cpr 1.16 0.18*** 0.24 0.13 2.92 1.05*** 1.18 0.25*** 7.69 2.25*** 4.33 0.85***Days’ work ent. 0.00 0.00 0.00 0.00 0.19 1.43** 0.62 0.43 1.07 5.76*** 0.23 2.84*Days’ work wage 0.26 2.10 0.00 0.56 0.68 1.81 0.14 0.70 9.85 21.50*** 8.08 18.36**



Table 12.5 Characteristics of different cohorts by gender of children, 2008–14

6–9-year-olds 10–14-year-olds 15–18-year-olds

2008 2014 2008 2014 2008 2014

Girls Boys Girls Boys Girls Boys Girls Boys Girls Boys Girls Boys

Sick in last 2 weeks 0.09 0.08 0.08 0.04* 0.07 0.06 0.03 0.03 0.06 0.08 0.04 0.02Attends school 0.67 0.71 0.79 0.72** 0.92 0.91 0.97 0.97 0.64 0.63 0.72 0.77Years of education 2.04 2.09 2.04 2.31** 5.79 5.75 5.78 6.04** 9.01 8.85 9.48 9.68Total days of work 7.40 5.13 1.60 2.12 18.63 24.00* 6.24 7.19 62.96 66.58 36.05 32.75**Days’ work ag. 6.24 2.33** 1.48 1.05 14.80 19.60* 5.13 5.19 39.77 37.14 17.38 13.10Days’ work cpr 0.44 0.58 0.12 0.20 1.52 1.74 0.70 0.34** 4.42 3.11* 1.85 1.99Days’ work ent. 0.00 0.00 0.00 0.00 0.53 1.55** 0.15 0.85* 3.64 5.31 2.49 1.60Days’ work wage 0.72 2.21 0.00 0.88 1.82 1.11 0.27 0.82 15.67 21.02 14.34 16.07



among non-Kinh suggests that convergence between the living standards ofthe two groups may take some time to realize.There are even more notable differences in child labour outcomes for

children of non-Kinh and Kinh descent, particularly in older age groups. In2014, ethnic minority children in the 10–14-year-old age group work onaverage 13.6 days outside of the home while children of Kinh householdswork only 3.8 days on average. Differences are most notable in agriculturalwork. For example, in 2014, children aged 10–14 of ethnic minority house-holds worked on average 11.6 days in the previous year in agricultural activ-ities (down from 26.8 in 2008). This is compared with 2.4 days on average forchildren of the same age from Kinh households. Among the 15–18-year-oldage group, children of ethnic minority households worked on average 30 daysin agricultural activities compared with only 8.7 for non-Kinh children of thesame age. Kinh children in this age group do, however, spend more daysworking for a wage (18.4) than non-Kinh (8.08). Overall, while welfare out-comes have improved for all children the gainsmade have not been enough toclose the large gap in welfare between children of ethnic minority householdscompared with those of Kinh descent. This is particularly the case for childrenover 10, with the biggest gaps apparent in the 15–18-year-old age group.

12.4 Panel Study

In this section we attempt to identify the key household characteristics thatare related to differences in the welfare outcomes of children. For this analysiswe create a household panel from 2008 to 2014, which tracks each childpresent in each household in 2008 through each round of the survey up to2014. For each welfare outcome we estimate the following model:

weliht ¼ βXht þ δ1femaleiht þ δ2ageiht þ αh þ τt þ εiht ð1Þwhere weliht is the welfare measure for child i in household h in time t; Xh is avector of household specific variables including characteristics of the house-hold head, income, land ownership, migration status of the household, thepresence of a household enterprise, and the incidence of natural and eco-nomic income shocks; female is a dummy indicator for whether the child isfemale; age is the age in years of the child; αh are household fixed effects; τt aretime dummies; and εiht is a statistical noise term.This model allows us to explore both individual and households factors that

are related to the welfare of children. The inclusion of household fixed effectscontrols for all time invariant household specific factors, such as, for example,ethnicity, geographical location, and other unobservable factors impacting onchild welfare. The time dummies control for any macroeconomic changes


248

over time affecting all children equally. As such we are analysing the withinhousehold variation in children’s outcomes across time. The vector of house-hold variables included in Xht allows us to disentangle the household specificfactors that are related to the welfare of children, although care should betaken in inferring any causality from these results. The coefficient δ1 will tell usthe extent to which welfare outcomes are better or worse for girls comparedwith boys in the same household. The inclusion of the age of each child willcontrol for the fact that welfare outcomes vary across age group, as was evidentfrom the cohort analysis presented in Section 12.3.

We focus on five main welfare indicators: (i) whether the child attendsschool; (ii) the years of education of the child; (iii) the total number of daysthe child worked outside of the home; (iv) the total number of days the childworked in agriculture in the last year; and (v) the total number of days thechild worked for a wage. The results are presented in Table 12.7.

We first consider the full sample of children aged 6 to 18. A number ofhousehold characteristics are found to be correlated with child welfare out-comes. Children with older heads of household are more likely to attendschool and spend fewer days working outside of the household. This is dueto fewer days spent in waged employment (column 5). Having a head ofhousehold with higher level education (more than second level education) ispositively correlated with the child attending school. A negative correlation isobserved between household income and the probability that children attendschool. In larger households children are less likely to attend school and havefewer years of education.

Children in higher income households spend more days working outside ofthe household (column 3), particularly in waged work (column 5). This sug-gests that in higher income households, children play a role in supportinghousehold income through working. This, however, may come at the expenseof children not attending school given the negative association foundbetween income and school attendance.

There is very little evidence that the assets of the household impact onwelfare outcomes. Basu, Das, and Dutta (2010) suggest that the relationshipbetween child labour and land holding may not be linear, but resemble aninverted-U relationship. We do find that the number of days worked inagriculture increases as the land size increases, but at a decreasing rate. How-ever, in the case of Viet Nam, it seems that the turning point is at extremelyhigh values of land holdings. Therefore, we can conclude that the relationshipbetween child labour in agriculture and land holdings is non-linear and on thepositive-sloped side of the reversed-U relationship. The opposite relationshipemerges when we consider the number of days worked for wage: the larger theland holdings, the less likely children are involved in waged work, but at adecreasing rate.

Children and Youths

249

A positive association is found between the ownership of durable goods(a measure of household assets) and the years of educational attainment ofchildren. Similarly, children have more years of education in households thathave an LUC. Both are suggestive of some positive correlation between wealthand educational investments in children.

Table 12.7 Panel data analysis of determinants of child welfare, 2008–14, 6–18-year-olds

(1) (2) (3) (4) (5)

Attendsschool

Years ofeducation

Daysworked

Days workedagriculture

Days workedwage

Household characteristics:Age 0.018*** 0.024 �0.469 0.061 �0.405

(0.006) (0.024) (0.909) (0.582) (0.622)Age2 �0.000 �0.000 0.002 �0.002 0.002

(0.000) (0.000) (0.010) (0.007) (0.007)Married 0.017 0.248 �1.115 3.505 �4.266

(0.043) (0.171) (4.902) (3.085) (3.484)Female �0.020 0.077 0.240 1.335 �2.575

(0.051) (0.215) (5.334) (3.182) (4.214)Higher ed. 0.040* �0.047 �2.668 �2.984 0.412

(0.022) (0.082) (2.711) (2.239) (1.567)HH size �0.024*** �0.081** 0.860 �0.773 1.552*

(0.008) (0.038) (1.081) (0.764) (0.801)Income �0.024*** �0.036 6.188*** �0.088 5.001***

(0.008) (0.036) (1.252) (0.808) (0.872)Loans �0.001 �0.005 �0.015 �0.033 0.014

(0.001) (0.005) (0.150) (0.100) (0.106)Land area �0.000 �0.000 �0.103 0.367* �0.435***

(0.002) (0.007) (0.261) (0.212) (0.144)Land area squared �0.000 0.000 �0.002 �0.003*** 0.002***

(0.000) (0.000) (0.001) (0.001) (0.001)Household enterprise 0.017 0.039 �1.514 �1.696 �4.150**

(0.016) (0.061) (2.287) (1.397) (1.663)Durables 0.003 0.055*** 0.305 0.198 0.243

(0.004) (0.021) (0.711) (0.439) (0.499)Red book 0.027 0.232*** 0.658 1.179 �1.388

(0.020) (0.079) (2.741) (1.867) (1.693)Natural shock 0.017 0.066 2.419 0.537 2.230**

(0.011) (0.052) (1.586) (1.107) (1.133)Economic shock 0.007 �0.069 2.770 3.139** �0.653

(0.013) (0.053) (1.944) (1.291) (1.409)

Child characteristics:Female �0.008 �0.018 3.486** 2.329** 0.701

(0.013) (0.066) (1.731) (1.053) (1.281)Age �0.007*** 0.765*** 5.946*** 2.921*** 2.398***


Notes: Each model includes household and time fixed effects. Robust standard errors clustered at the household level inparentheses. ***p<0.01, **p<0.05, *p<0.1.



250

Following Edmonds and Turk (2004), we also include a dummy variable thatcaptures whether the household runs an enterprise. While we do not find anyimpact of household enterprises on education, we do provide evidence thatthe number of days worked for wage is lower when a household enterprise ispresent. It is indeed likely that children are employed in the householdenterprise rather than working outside the household.

In households that experience natural shocks (floods, droughts, pest infest-ations), children spend more days working in waged employment while inhouseholds that experience economic shocks (illness, unemployment, shocksto input or crop prices, etc.), children spendmore days working in agriculturalactivities. Both of these results suggest that households use child labour as ashock-coping mechanism. In the case of natural shocks, children are put towork in waged employment, given that natural shocks usually affect theagricultural activities of households. In the case of economic shocks, childresources are diverted into agriculture, perhaps to enable other householdmembers to enter waged employment or work in household enterprises.

Our panel data results confirm our findings from the cohort analysis thatthere are differences in the welfare outcomes of boys and girls. In particular,we find that controlling for the age of the children, girls aremore likely to haveexperienced sickness in the previous two weeks than boys and girls spendmore days working outside of the home. In particular, girls spend more daysengaged in agricultural activities than boys.

In the next step of our analysis we focus specifically on households thatinclude children aged 10–15, given that they are the most vulnerable in termsof exposure to child labour and consequentially negative impacts on educa-tion outcomes. We estimate the regression model as in equation (1) for thesame set of welfare outcomes. The results are presented in Table 12.8.

Fewer of the household characteristics are statistically significant once thesample is reduced to 10–15-year-olds. We find that children are less likely toattend school and have fewer years of schooling in larger households. They arealso more likely to work for a wage outside of the home. Children in higherincome households also spend more days working, particularly in wagedemployment, suggesting that there are cases where household income isbeing supported by child labour. Exposure to both natural and economicshocks also increases the number of days that children aged 10–15 spendworking outside of the household, particularly in agriculture.

As highlighted in Section 12.1, there is a large literature which suggests thatfemale empowerment, and in particular an increase in the resources held in thehands of women, is beneficial for children. An increase in female bargainingpower within the household is expected to decrease child labour, especiallyamong girls (see, for example, Duflo (2003) and Qian (2008)). To explore thispossibility in the Vietnamese case we consider two indicator variables for the

Children and Youths

251

empowerment of women within the household: (i) an indicator variable forwhether a woman manages the land owned by the household; and (ii) theproportion of total days worked by women for a wage. The latter is consideredan indicator of female empowerment on the basis that income earned throughwaged employment is more likely to be under the control of the person whoearned the income. We include each of these indicators in the regression

Table 12.8 Panel data analysis of determinants of child welfare, 2008–14, 10–15-year-olds

(1) (2) (3) (4) (5)

Attendsschool

Years ofeducation

Daysworked


Days workedwage

Household characteristics:Age 0.023*** 0.013 �0.598 0.013 �0.236

(0.008) (0.031) (1.162) (0.843) (0.578)Age2 �0.000*** �0.000 0.010 0.001 0.002

(0.000) (0.000) (0.013) (0.010) (0.006)Married �0.009 0.007 �6.550 �3.858 �0.822

(0.063) (0.138) (6.981) (6.351) (3.180)Female �0.033 �0.104 �8.124 �3.114 �5.870

(0.065) (0.177) (8.008) (5.507) (5.341)Higher ed. �0.018 �0.063 �0.413 �0.800 0.389

(0.024) (0.120) (5.283) (4.361) (0.512)HH size �0.043*** �0.097* 0.702 �1.106 1.383**

(0.012) (0.051) (1.513) (1.305) (0.607)Income �0.012 �0.024 2.592* �0.439 2.163***

(0.008) (0.039) (1.449) (1.139) (0.763)Loans 0.000 �0.005 0.130 0.138 0.039

(0.001) (0.006) (0.189) (0.140) (0.107)Land area �0.004 �0.010 0.032 0.136 �0.093

(0.003) (0.012) (0.451) (0.428) (0.107)Land area squared 0.000 0.000 �0.001 �0.002 0.001

(0.000) (0.000) (0.004) (0.004) (0.001)Household enterprise 0.007 �0.022 1.280 �1.545 �1.284

(0.015) (0.078) (2.831) (2.082) (1.322)Durables 0.006 0.078*** 1.234 0.848 0.404

(0.005) (0.030) (0.865) (0.542) (0.616)Red book 0.022 0.151 �1.614 �1.614 �0.605

(0.021) (0.100) (3.360) (2.699) (1.608)Natural shock 0.011 0.024 4.218** 2.716* 1.160

(0.012) (0.064) (1.861) (1.527) (0.894)Economic shock 0.023* 0.094 2.119 3.570* �1.381

(0.012) (0.062) (2.549) (2.036) (1.204)

Child characteristics:Female �0.003 �0.010 0.306 0.393 0.740

(0.015) (0.083) (2.114) (1.586) (1.152)Age �0.028*** 0.842*** 6.183*** 4.208*** 1.377***


Notes: Each model includes household and time fixed effects. Robust standard errors clustered at the household level inparentheses. ***p<0.01, **p<0.05, *p<0.1.



252

models. The results are presented in Table 12.9. Only results for the empower-ment variables are presented for ease of illustration but eachmodel includes thefull set of household and individual control variables.

There is some evidence to suggest that in households where a woman isresponsible for managing the land owned by the household, children aremore likely to attend school, although it appears that they are also more likelyto work more for a wage. It is also the case that, in households where womenspend a greater proportion of their time working for a wage, as opposed toother types of activities, children work significantly fewer days and, in par-ticular, work significantly fewer days in agricultural activities.

12.5 Conclusions

This chapter investigated how the lives of the children and youths living inrural Viet Nam have been affected by the significant structural transformationexperienced in Viet Nam over the last decade. We analyse different aspects ofchild wellbeing: health, education attendance and attainment, and engage-ment in labour (agricultural, household enterprise, and waged employment).The analysis depicts a society that has made great progress towards improvingchild welfare. Over the span of six years, the health of children and youngpeople has improved. School attendance has also increased, in particular forchildren above the age of 10. This is particularly notable given that this agegroup is past the age of compulsory primary school.We also observe a decreasein child labour, which is evenmore notable for themost vulnerable age group,the young cohort.

Table 12.9 Panel data analysis of female empowerment and child welfare, 2008–14,10–15-year-olds

(1) (2) (3) (4) (5)

Attendsschool

Years ofeducation

Daysworked


Days workedwage

Empowerment IndicatorsFemale manager 0.032** 0.086 0.129 �1.523 1.056**

(0.012) (0.072) (2.155) (1.917) (0.537)Proportion total days worked by

women that are spent in wagedemployment

0.005 �0.038 �9.530** �9.508*** 1.267(0.020) (0.115) (3.912) (3.052) (1.837)

Observations 3,427 3,427 3,428 3,428 3,428Number of HH 1,064 1,064 1,064 1,064 1,064

Notes: Each model includes household and time fixed effects and the full set of household and individual characteristicsincluded in Table 12.8. Robust standard errors clustered at the household level in parentheses. ***p<0.01, **p<0.05,* p<0.1.


Children and Youths

253

Many challenges, however, still lie ahead. While both girls and boys haveexperienced improvements in health and schooling outcomes, we find thatboys benefited more than girls. Similarly, while wellbeing has increasedover time for both minority and non-minority groups, our analysis high-lights the fact that a substantial difference in the level of welfare still remainsbetween the two groups. Of particular concern is the widening gap ineducational outcomes. With slower rates of human capital accumulationfor the poorest groups in society, convergence in living standards will bemore difficult and will take a longer time to attain. Indeed, the World Bank(2016) highlights inequalities in opportunities for ethnic minority childrenand stresses the importance of expanding government initiatives to addressthese inequalities.Nevertheless, the large gains in the welfare of children in Viet Nam over the

last eight years is a strong signal that structural transformation is paving theway for a better standard of living for the next generation and future gener-ations to come.

References

Basu, K., S. Das, and B. Dutta (2010). ‘Child Labour and HouseholdWealth: Theory andEmpirical Evidence of an Inverted-U’. Journal of Development Economics, 91(1): 8–14.

Beegle, K., R. Dehejia, and R. Gatti (2009). ‘Why Should We Care about Child Labour?The Education, Labour Market, and Health Consequences of Child Labour’. Journal ofHuman Resources, 44(4): 871–89.

Duflo, E. (2003). ‘Grandmothers and Granddaughters: Old Age Pension and Intra-Household Allocation in South Africa’. World Bank Economic Review, 17(1): 1–25.

Edmonds, E. (2005). ‘Does Child Labour Decline with Improving Economic Status?’Journal of Human Resources, 40(1): 77–99.

Edmonds, E. and N. Pavcnik (2005a). ‘Child Labour in the Global Economy’. Journal ofEconomic Perspectives, 19(1): 199–220.

Edmonds, E. and N. Pavcnik (2005b). ‘The Effect of Trade Liberalization on ChildLabour’. Journal of International Economics, 65(2): 401–19.

Edmonds, E. and C. Turk (2004). ‘Child Labour in Transition in Viet Nam’. InP. Glewwe, N. Agrawal, and D. Dollar (eds), Economic Growth, Poverty and HouseholdWelfare in Viet Nam. Washington DC: World Bank.

Krueger, A. B. (1997). ‘International Labour Standards and Trade’. In M. Bruno, andB. Pleskovic (eds), Annual World Bank Conference on Development Economics, 1996.Washington DC: World Bank.

Newman, C. (2015).Gender Inequality and the Empowerment of Women in Rural Viet Nam.WIDER Working Paper 2015/066. Helsinki: UNU-WIDER.

O’Donnell, O., F. Rosati, and E. van Doorslaer (2005). ‘Health Effects of Child Work:Evidence from Rural Viet Nam’. Journal of Population Economics, 18(3): 437–67.


254

Pitt, M. and S. R. Khandker (1998). ‘The Impact of Group-Based Credit Programs onpoor Households in Bangladesh: Does the Gender of Participants Matter?’ Journal ofPolitical Economy, 106(5): 958–96.

Qian, N. (2008). ‘Missing Women and the Price of Tea in China: The Effect of Sex-Specific Earnings on Sex Imbalance’. Quarterly Journal of Economics, 123(3): 1251–85.

World Bank (2016). Vietnam 2035: Toward Prosperity, Creativity, Equity and Accountability.Washington DC: World Bank.

Children and Youths

255

13

Ethnic Disadvantage

Evidence Using Panel Data

Saurabh Singhal and Ulrik Beck

13.1 Introduction

Viet Nam is an ethnically diverse country with fifty-four officially recognizedethnic groups. The Kinh, the ethnic majority, constitute about 86 per cent ofthe population. Among the non-Kinh, the largest ethnic groups are the Tay,Thai, Muong, and Khmer, who account for a little less than 2 per cent of thepopulation each (World Bank 2009).While Viet Nam has witnessed rapid growth and poverty decline since the

Doi Moi reforms initiated in 1986, qualitative and quantitative evidenceindicates that these gains have not been shared equally across ethnic groups(World Bank 2012; The Economist 2015). Using household income as anindicator for welfare, research has found not only that the non-Kinh weresystematically worse off than the Kinh but also that this gap widened duringthe 1990s (Van de Walle and Gunewardena 2001; Baulch et al. 2007; Baulch,Pham, and Reilly, 2012) and the likelihood of them escaping poverty wasrelatively much smaller (Glewwe, Gragnolati, and Zaman 2002).1 A variety ofexplanations have been put forward for the poor performance of the minorityhouseholds in Viet Nam. The ethnic minorities are less endowed (in keyaspects such as land holdings, education, access to credit, etc.) and also facelower returns to endowments. While the remote location of the minorityhouseholds can partially explain the gap in endowments, research has con-sistently found that it is not the sole reason for the gap.

1 Similarly, Pham and Reilly (2009) find significant ethnic wage gaps in the labour market inViet Nam.

In this study, we use the Viet Nam Access to Resources Household Survey(VARHS) to examine how the welfare of ethnic minorities in Viet Nam hasevolved over the period 2006–14. Specifically, we check whether the ethnicgaps still exist, and if so, whether there has been any convergence over time; wealso examine the factors constraining the growth ofminority households. In linewith the existing literature, wefind that the non-Kinh continue to lag behindnotonly in terms of income and consumption, but also on a variety of other indica-tors of living standards such as access to water and toilets. An examination of thehousehold income structure reveals that while the Kinh are more likely to diver-sify into wage employment and non-farm household enterprises, the non-Kinhrely more heavily on common pool resources (CPR). We explore the constraintsto growth and income diversification and find several differences that can helpexplain thewelfare gap. The quality of agricultural land and ownership certificaterates are lower for non-Kinh households, and the effects of these persist evenwhen we control for the fact that non-Kinh households tend to live in certainprovinces. Non-Kinh households also experience more problems producing andselling their agricultural output andhaveworse access to credit.Whilehistoricallynon-Kinh households have been more remotely located, their relative isolationappears to have abated over time. On the other hand, we find evidence ofsegmentation in social networks along ethnic lines. In the last section, we exploitthe richness of the VARHS data to examine differences among groups thatconstitute the non-Kinh andfind significant heterogeneitywithin the non-Kinh.

The VARHS data allows us to classify households into the various ethnicgroups based on the ethnicity of the household head. In this study, a house-hold is defined as a Kinh household if the household head belongs to the Kinhethnicity.2 Among the minorities, studies typically club the Hoa (or theChinese) along with the Kinh as the Hoa are economically at least as well offas the Kinh. In this study, we consider the Hoa along with the non-Kinh as weonly observe four Hoa households in the VARHS data.

Before proceeding to the analysis, we first present a picture of ethnic differ-ences in key welfare indicators in contemporary Viet Nam. Table 13.1 presentsbasic demographic information and household characteristics by ethnicityusing the 2014 VARHS data. On average, the non-Kinh are significantly morelikely tohave ahouseholdheadwho is illiterate and a larger household, and theaverage monthly per capita food expenditure and income of the non-Kinh isnearly half of that of the Kinh. As shown in Chapter 2, another key definingcharacteristic of the non-Kinh is that they are geographically concentrated inthe mountainous Northern region and the Central Highlands. The remainderof this chapter explores the sources and trends of these gaps in greater detail.

2 In some cases, the ethnicity of other household members may differ due to inter-marriages.Unfortunately, we are unable to examine such cases.

Ethnic Disadvantage

257

13.2 Welfare Levels and Trends

Figure 13.1 shows the evolution of mean real food expenditures and realincome per capita for Kinh and non-Kinh households from 2006 to 2014,along with 95 per cent confidence intervals. While food expenditures of bothKinh and non-Kinh households increased significantly over the period, thelevel of food expenditures for minority households was significantly lowerover the entire period. There are no signs of minority households catching upto the expenditure levels of their Kinh counterparts as growth rates have beenalmost the same over the period: from 2006 to 2014, food expendituresof Kinh households increased by 53.5 per cent (or 5.5 per cent annually)whereas those of non-Kinh households increased by 56.4 per cent (5.7 percent annually).The income time series presents a similar picture.3 For both Kinh and non-

Kinh households, income per capita increased over the period 2008–14 byannual rates of 7.8 and 8.5 per cent, respectively.While the raw average growth rates may be similar for the two groups, there

is no evidence of convergence and the average income of non-Kinh house-holds in 2014 was just half of themean Kinh income. To illustrate, if one takesthe difference in income in 2014 as a point of departure and projects futureKinh and non-Kinh mean income using the annual growth rates of the2008–14 period, it would take 104 years before non-Kinh households caughtup with their Kinh counterparts. It is, of course, unrealistic to project currentgrowth ratesmore than 100 years into the future, but it does illustrate the need

Table 13.1 Descriptive statistics by ethnicity for 2014

Kinh Non-Kinh Difference

Household head illiterate (%) 4.35 31.24 �26.9***Household head female (%) 27.06 11.89 15.17***Household size (mean) 3.9 5.05 �1.15***Food expenditure, monthly per capita 499.07 283.44 215.63***Income, monthly per capita 2313.53 1174.87 1138.66***Region of residence (%)Central Highlands 11.54 20.28Mekong River Delta 15.98 0North 16.39 71.33Red River Delta 26.89 0.93Central Coast 29.2 7.46Number of households 1733 429

Notes: Food expenditure and Income are in real 1000 VND. *** indicates significance at the 99%confidence level.

Source: Authors’ calculations based on the VARHS database.

3 Comparable income estimates can only be constructed for the period 2008–14.


258

to focus more on the minority ethnicities moving forward if the expenditureand income gaps are to be closed.

To summarize, non-Kinh households were worse off than Kinh householdsover the entire period. This was caused by the combination of lower initiallevels of income and food expenditure, and similar growth rates for the twogroups. A logical next question is whether income evolved differentially forthese two groups if one more directly compares households with the sameinitial levels of income. This can be done by exploiting the panel dimension ofthe VARHS database. Figure 13.2 presents non-parametric regression estimatesof real monthly per capita income in 2014 on real per capita income in 2008separately for Kinh and non-Kinh households. The solid lines show the aver-age income level in 2014 for a given level of income in 2008. A striking pictureemerges: for a wide range of initial incomes, Kinh households experiencedhigher income growth over the period. For example, non-Kinh householdswho earned around 500,000 VND per capita in 2008 had, on average, almostdoubled their income in 2014 to around 1,000,000 VND per capita. However,Kinh households who earned a similar amount in 2008 could have expectedto triple their income to 1,500,000 VND per capita in 2014.4 Similarly, results

0

100

200

300

400

500

600

2006 2008 2010 2012 2014

Rea

l 100

0 V

ND

Year

(a) Food Expenditure

0

500

1000

1500

2000

2500

3000

2008 2010 2012 2014

Rea

l 100

0 V

ND

Year

(b) Income

Kinh Non-KinhKinh Non-Kinh

Figure 13.1 Evolution of monthly food expenditure (a) and income (b) by ethnicity inreal 1000 VND, 2006–14Notes: Dashed lines represent 95% confidence intervals. Income data is not available in 2006.Expenditure and income are represented in real June 2014 prices.Source: Authors’ calculations based on the VARHS database.

4 We note that the household-specific growth rates of income one gets from these examplehouseholds are much higher than average income growth rates. This is not unusual in this type ofsetup and is caused by negative idiosyncratic shocks in 2008. These shocks suppress incomes in2008 but are gone by 2014. Therefore, the income growth for these households seems very high.

Ethnic Disadvantage

259

from regression analysis in Chapter 10 indicate that, controlling for pastendowments and income, the income growth rate of non-Kinh householdsis significantly lesser than that of the Kinh households.Welfare is not exclusively determined by monetary indicators such as

income and expenditure. Figure 13.3 shows the evolution of a series of assetindicators by ethnicity. Figure 13.3 (a–c) details the evolution of ownership ofcows, buffaloes, and pigs that are all used in agricultural production. Perhapssurprisingly, in the study period, non-Kinh households did better in terms ofthe number of pigs and buffaloes and were on par with Kinh households interms of the number of cows. How is this connected to the clear expenditureand income discrepancy in favour of Kinh households? One possibility is thatas agriculture becomes increasingly mechanized, draft animals such as cowsand buffaloes become less important. The process of mechanization takesplace at a more rapid pace for richer households since they have the requisiteeducation, capital, and credit access. Since Kinh households are in generalbetter off, they aremore able to implementmodern agricultural methods. Thisexplanation is consistent with the general decline in the number of cows andbuffaloes observed in Figure 13.3 (a and b, respectively). Another possibility isthat non-Kinh households with worse access to credit are more likely to utilize

0

1000

2000

3000

4000

5000

6000

7000

0 500 1000 1500 2000 2500 3000

Per

cap

ita

mo

nth

ly in

com

e, 2

014,

'000

VN

D

Per capita monthly income, 2008, '000VND

Kinh Non-Kinh

Figure 13.2 Non-parametric estimates of Kinh and non-Kinh income growth, depend-ing on initial income, 2008 and 2014Notes: Dashed lines represent 95 per cent confidence intervals. In order to increase legibility, thex-axis is cut off at 90,000 VND, which is above the 95th percentile of 2008 incomes. All values arein June 2014 prices.Source: Authors’ calculations based on the VARHS database.


260

animals as a store of value to be realized in the event of a negative incomeshock. The issue of access to credit is discussed in more detail in Section 13.4.

Figure 13.3 (d–f) details the evolution of ownership of three durable con-sumption assets: colour TVs, motorcycles, and bicycles. Here, the trend iscloser to the evolution of monetary indicators: for both Kinh and non-Kinhhouseholds, ownership rates of colour TVs and motorcycles increased, but thelevel of non-Kinh ownership lagged behind throughout the entire period.Ownership rates of bicycles fell for both groups, most likely because of substi-tution by motorcycles or, more rarely, cars.

Figure 13.3 (g–i) shows three housing indicators: toilets, water supply, and areaof house in square metres, respectively. Over the entire period 2006–14, Kinhhouseholds improved their outcomes in all three dimensions. In 2014, over 90per cent had access to an improved water supply such as tap or well water and animproved toilet facility such as a flush, squat, or double-vault compost toilet.Houses became larger as well: in the span of eight years, the average house sizeincreased by almost 40 per cent. For non-Kinh households, the picture is bleaker:in 2014, less than 60 per cent had a good toilet and less than 50 per cent hadaccess to good water supply. The steady improvement in monetary welfare isreflected in the housing indicators only for Kinh households: worryingly, for

00.10.20.30.40.50.6

2006 2008 2010Year

(a) No. of cows

2012 20140

0.5

1

1.5

2

2006 2008 2010 2012 2014Year

(b) No. of buffalos

0

0.5

1

1.5

2006 2008 2010Year

(e) No. of motorcyles

2012 20140

0.20.40.60.8

11.2

2006 2008Year2010 2012 2014

(d) No. of color TVs

00.20.40.60.8

1

2006 2008 2010 2012 2014

(g) Access to good toilet

YearKinh Non-Kinh

00.20.40.60.8

1

2006 2008 2010 2012 2014

(h) Access to good water

YearKinh Non-Kinh

00.5

11.5

22.5

33.5

2006 2008 2010 2012 2014

(c) No. of pigs

Year

00.20.40.60.8

11.21.4

2006 2008 2010 2012 2014

(f) No. of bicycles

Year

405060708090

100

2006 2008 2010 2012 2014

Kinh Non-Kinh

(i) Area of house (sqm)

Year

Figure 13.3 Household asset ownership rates by asset and ethnicity, 2006–14Notes: A good toilet is defined as flush, squat, or a double-vault compost toilet. A good water supplyis defined as tap or well water.Source: Authors’ calculations based on the VARHS database.

Ethnic Disadvantage

261

non-Kinh households, the proportion of households with a good toilet fellbetween 2006 and 2010 and the proportion with access to improved watersupply fell from 2006 to 2012. Non-Kinh households were more likely to ownamotorbike or a colour TV than to have access to an improvedwater supply or toa good toilet facility in 2014. Finally, the housing size picture is more optimisticin that the average area increased for both groups. What is less optimistic is thewidening of the gap between Kinh and non-Kinh households over the studyperiod.While the square-metre area increased for both groups, it increasedmuchmore slowly for non-Kinh households.Finally, welfare may also be assessed in terms of the educational attainment

of children. As discussed in Chapter 12 and the working paper version of thischapter (Singhal and Beck 2015), there are stark persistent ethnic gaps inschool attendance and educational attainment (as measured by grade forage) among older age cohorts. This is particularly the case after 15 years ofage, which corresponds to the end of junior high school. A possible explan-ation is a greater reliance on child and adolescent labour as a risk-copingmechanism among minority households (Beck, Singhal, and Tarp 2016).

13.3 Structure of Household Income

In order to better understand the observed lack of convergence between Kinhand non-Kinh households, we now explore how the patterns of economicactivity differ between the Kinh and the non-Kinh. Is the likelihood of diversi-fying out of agriculture into wage employment, household enterprises, or CPRdifferent for the two groups? Income diversification is important as it allowshouseholds to weather shocks, smooth consumption, and boost income. Forthe case of rural Viet Nam, Khai and colleagues (2013) show that incomediversification over the period 2008–12 led to an increase in household welfare.Similarly, Oostendorp, Trung, and Tung (2009) find that operating non-farmhousehold enterprises significantly increased household income in Viet Namduring 1993–2002 (for a broader discussion, see Chapter 5). We first examineethnic differences in such diversification using the 2014 data.We begin by splitting the sample into diversifiers and non-diversifiers, that

is, those who solely depend on agriculture for their income and those whohave at least one other non-agricultural source of income. The first row ofTable 13.2 shows the proportion of Kinh and non-Kinh households that arenon-diversifiers. The non-Kinh are more likely to have diversified out ofagriculture in 2014: 13 per cent of Kinh households depend solely on agricul-ture as opposed to 7.7 per cent of non-Kinh households.While almost all the households rely on agriculture to some degree, they

also derive income from wage employment, household non-farm enterprises,


262

and CPR.We further categorize the diversifying households into the followingmutually exclusive groups: those that combine agriculture with: (i) wageemployment; (ii) household enterprises; (iii) CPR; (iv) wage employmentand CPR; (v) wage employment, household enterprises, and CPR; or (vi)some other combination. Looking at differences across ethnicity for eachcategory in Table 13.2, we find that the Kinh are more likely to diversify intoeither wage employment or household enterprises, while the non-Kinh aremore likely to depend on CPR, either independently or in conjunction withwage employment or household enterprises. Conditional on diversifying outof agriculture, the 2014 data reveals that the non-Kinh and the Kinh differsignificantly on the income-generating activities they diversify into.

Next, we move beyond this static view to examine whether patterns ofdiversification strategies have changed over time. Figure 13.4 (a–c) shows theproportion of Kinh and non-Kinh households that derived income fromhousehold enterprises, wage employment, and CPR, respectively, over theperiod 2006–14. Figure 13.4 (a) shows that while the proportion of Kinhhouseholds involved in household enterprises declined slightly over timefrom 0.38 in 2006 to 0.27 in 2014, it is consistently more than that of thenon-Kinh. More importantly, we find large fluctuations in the proportion ofnon-Kinh households engaged in household enterprises. This flux in and outof self-employment indicates that the household enterprises operated by non-Kinh households are transitory and not able to survive for long.

Figure 13.4 (b) reveals somewhat similar dynamics with respect to wageemployment. While the Kinh and non-Kinh were equally likely to engage inwage employment in 2006 and 2014, the non-Kinh exhibited more variabil-ity. The picture is completely different when we examine the trends for CPR inFigure 13.4 (c). While the proportion of Kinh households using CPR increased

Table 13.2 Income diversification by ethnicity in 2014

Kinh Non-Kinh Difference

Non-diversifiersAgriculture only 12.98 7.69 5.29***DiversifiersAgriculture and wage only 29.20 14.22 14.98***Agriculture and business only 7.44 0.93 6.51***Agriculture and CPR only 5.83 16.78 �10.95***Agriculture, wage, and CPR 14.25 45.92 �31.67***Agriculture, wage, business, and CPR 2.14 6.53 �4.39***Other combinations 28.16 7.93 20.23***Observations 1,733 429

Notes: The last column reports the t-test of proportions. *** indicates significance at the 99%confidence level.

Source: Authors’ calculations based on the VARHS database.

Ethnic Disadvantage

263

modestly over 2006–14, the proportion of non-Kinh more than doubled from36 per cent in 2006 to 75 per cent in 2014.In this section, we find that the structure of household income varies

significantly between the Kinh and the non-Kinh. As noted earlier in thissection, diversification out of agriculture is positively associated with welfare.Since the non-Kinh more likely to diversify (less likely to be solely dependenton agriculture), it begs the question: why does the ethnic gap in incomepersist? We examine three avenues of diversification: wage employment,household enterprises, and CPR, and find that non-Kinh households thatdiversify are more likely to depend on CPR as opposed to Kinh householdsthat rely primarily on wage employment and household enterprises. Thisindicates that the avenue of diversification matters: despite diversificationout of agriculture, a possible explanation for the poor performance of minor-ity households is the low returns from CPR. Furthermore, as income from CPRis more susceptible to climate change, this finding indicates severe implica-tions for the vulnerability of non-Kinh households in the future.

13.4 Constraints to Agricultural and Non-Agricultural Production

We now turn to identifying some of the constraints on agricultural growthand the ability to diversify out of agriculture, as established in Section 13.3.We do this by looking at differences in plot characteristics, reported problemsregarding agriculture, credit access, remoteness, and social networks.

13.4.1 Land and Agriculture

This section investigates how the agricultural production of non-Kinh house-holds is differentially constrained compared to their Kinh counterparts. This is

0

0.1

0.2

0.3

0.4

2006 2008 2010 2012 2014Year

(a) HH enterprise

Kinh Non-Kinh

0

0.2

0.4

0.6

0.8

2006 2008 2010 2012 2014Year

(b) Wage employment

Kinh Non-Kinh

0

0.2

0.4

0.6

0.8

1

2006 2008 2010 2012 2014Year

(c) CPR

Kinh Non-Kinh

Figure 13.4 Income diversification, 2006–14Note: The dummy variable ‘HH enterprise’ takes a value 1 if the household operates at least oneenterprise and 0 otherwise. Similarly, the variables ‘Wage employment’ and ‘CPR’ take a value 1 if ahousehold participates in off-farm wage employment and collects CPR, respectively.Source: Authors’ calculations based on the VARHS database.


264

done by analysing differences in land quality and ownership status and self-reported problems faced pre- and post-production.

Figure 13.5 shows the difference in some characteristics of land quality as wellas in red-book ownership between Kinh and non-Kinh households in 2014.These are calculated by regressing the outcome variable on a dummy variableequal to one if the household is of an ethnicity other than Kinh. As shown inTable 13.1, non-Kinh households are not equally distributed between the prov-inces. Rather, the non-Kinh tend to live in upland areas where climatic condi-tions such as rainfall and temperature, as well as soil fertility and composition,differ fundamentally from those in the lowland coastal areas. In order to ensurethat the differences observed are real differences between Kinh and non-Kinhfarmers, Figure 13.5 also includes estimates that are based only on differencesbetween Kinh and non-Kinh households within the same province. Formally,this is done by including province fixed effects in the regressions.

Non-Kinh farmers have to travel significantly longer distances to their plots,they have fewer plots with soil or rock bunds in place, and they have a largershare of plots without formal ownership rights in the form of a red book. Theeffects of these factors are all significant, although smaller, using the within-province estimates. On the other hand, non-Kinh farmers, on average, facefewer growing restrictions and have more terraces on their plots than their

–20

–10

0

10

20

30

40

Long distanceto plots

No growingrestrictions

No bunds No terraceesNo terracees No red book

Non-Kinh effect Within-province non-Kinh effect

Figure 13.5 Land quality and redbook ownership, 2014Notes: ‘Long distance to plots’ is defined as the share of plots that aremore than 1 km away from theresidence. Categories of ‘No growing restrictions’, ‘No bunds’, ‘No terraces’, and ‘No red book’ arethe share of plots on which there are no growing restrictions, no soil or rock bunds present, noterraces built, or where the household has no red book for the plot, respectively. All shares arecalculated as simple averages over the number of plots the household owns and operates or rentsin. Shares are reported in percentage. Black bars indicate 95% confidence intervals. Within-province non-Kinh effects are calculated by including province fixed effects in the regressions.Source: Authors’ calculations based on the VARHS database.

Ethnic Disadvantage

265

Kinh counterparts. These effects are still significant using the within-provinceestimator, although they are significantly smaller and the lower confidencebounds are very close to ‘0’. That this is the case for these two variables makesintuitive sense: there are fewer restrictions on having to grow rice in the uplandareas where rice production is of less importance, and this is where a higherproportion of the non-Kinh households live. Similarly, there are more terracesin the more hilly and mountainous upland areas. In sum, non-Kinh farmersface additional constraints in terms of access to their land, the quality of theland they own, and tenure security compared to their Kinh counterparts.We next turn to the problems that households face before harvest

(Figure 13.6) and after harvest (Figure 13.7). In both cases, Kinh householdsare more likely to report no problems than non-Kinh households. Forinstance, in 2014, 85 per cent of Kinh households reported facing no problemsbefore and after harvest. On the other hand, 69 per cent of non-Kinh farmersfaced no problems before harvest and 63 per cent faced no problems afterharvest. What is the nature of the problems faced before harvest? According toFigure 13.6, non-Kinh farmers are more likely to face a lack of suppliers, notbeing able to buy on credit, and to face poor transport infrastructure. Kinhfarmers, on the other hand, are increasingly impeded by lack of information, atrend not observed for non-Kinh households, possibly because they are facingother and more pressing problems. In 2008, almost 40 per cent of non-Kinhfarmers reported facing very high input prices; in 2014, this fell to around10 per cent for both Kinh and non-Kinh farmers. In terms of problems after

0

20

40

60

80

100

2008 2010 2012 2014Year

(a) No problems

0

5

10

15

2008 2010 2012 2014Year

(b) Lack of suppliers

0

10

20

30

40

50

2008 2010 2012 2014Year

(c) Prices too high

0

5

10

15

20

25

2008 2010 2012 2014Year

(d) Not enough information

Kinh Non-Kinh

02468

101214

2008 2010 2012 2014Year

(e) Not able to buy on credit

Kinh Non-Kinh

0

5

10

15

20

25

2008 2010 2012 2014Year

(f) Poor transport infrastructure

Kinh Non-Kinh

Figure 13.6 Most important constraints before harvest as reported by households byyear, 2008–14Notes: Some categories with very few answers are left out. Shares are reported in percentage.Source: Authors’ calculations based on the VARHS database.


266

harvest, more non-Kinh farmers are concerned about lack of output storage,information about prices, and high transportation costs, even though the lastseems to be of less importance in later years.

Overall, non-Kinh households face additional agricultural constraints dueto lower quality of plots, lower ownership rates, and more problems regardingagriculture both before and after harvest.

13.4.2 Credit and Borrowing

This section looks at differences in access to credit (both formal and informal)between Kinh and non-Kinh households. A loan is often required if a farmerwants to expand agricultural production or start a non-farm enterprise. Pooraccess to credit can therefore severely limit a household’s possibilities foragricultural growth and diversification out of agriculture.

Figure 13.8 shows some information on loans. Figure 13.8 (a), which showsthe share of households that borrowed money in the last two years, indicates

0

20

40

60

80

100

2008 2010 2012 2014

Year

(a) No problems

0

5

10

15

20

2008 2010 2012 2014

Year

(c) Lack of information about prices

Kinh Non-Kinh

0

5

10

15

20

25

2008 2010 2012 2014

Year

(d) High transportation costs

Kinh Non-Kinh

0

5

10

15

20

25

2008 2010 2012 2014Year

(b) Lack of output storage

Figure 13.7 Most important constraints after harvest as reported by households byyear, 2008–14Notes: Households were asked to list up to two problems. Some categories with very few answers areleft out. Shares are reported in percentage.Source: Authors’ calculations based on the VARHS database.

Ethnic Disadvantage

267

that, in the later part of the study period, and especially in 2014, a larger shareof non-Kinh households borrowed money. At first glance, this would indicatethat ethnic minority households do not have worse access to credit. However,Figure 13.8 (b) shows that more non-Kinh households have had their loanapplications rejected (although note that the overall rejection rate is low). Thisdiscrepancy is particularly large in 2014, during which almost 4 per cent ofnon-Kinh households had a loan rejected in the previous two years, whereasthis was the case for less than 1 per cent of Kinh households. It should benoted that borrowingmoney is not always a good thing: borrowing because ofdifficulties in making ends meet is very different from borrowing forinvestments.Figure 13.8 (c) looks into the types of loans in more detail by showing how

the amount borrowed varies between ethnicities. The average loan size fornon-Kinh households was less than half of the size of Kinh household loans in2008. This discrepancy increased over time: in 2014, the average non-Kinhloan size was reduced to less than a third of the size of the loans of Kinhhouseholds.To summarize, the picture is not as positive as a quick glance at Figure 13.8 (a)

seems to indicate: although more non-Kinh households got loans in the studyperiod, more non-Kinh households also had loan applications rejected and,when they did get a loan, it was substantially smaller. Most worryingly, thediscrepancy appears to have increased over time.

13.4.3 Remoteness

As mentioned in Section 13.1, the ethnic minorities of Viet Nam tend to livein more mountainous and remote parts of the country. Longer distances topopulation centres can result in less access to public services and infrastructureand increased transportation costs. Over the years, the government has

0102030405060

2008 2010 2012 2014

Perc

enta

ge

Year

(a) HH borrowed money

Kinh Non-Kinh

0

1

2

3

4

2008 2010 2012 2014

Perc

enta

ge

Year

(b) HH had loan rejected

Kinh Non-Kinh

0

20

40

60

80

2008 2010 2012 2014

Mill

ion

s o

f V

ND

Year

(c) Amount borrowed

Kinh Non-Kinh

Figure 13.8 Access to credit by ethnicity, 2008–14Notes: Shares are reported in percentage. The figures in (c) are calculated conditional on receiving aloan and are reported in million Vietnamese Dong. ‘HH’ is abbreviation for household.Source: Authors’ calculations based on the VARHS database.


268

targeted several programmes such as the Socio-Economic Development Pro-gram for the Communes Facing Greatest Hardships in the Ethnic Minorityand Mountainous Areas (‘Program 135’ or ‘P135’) to support infrastructuredevelopment and public services in such areas.5

Although there have not been any rigorous evaluations of such policies, weexamine whether minority households continue to systematically differ intheir access to infrastructure due to their geographical location. In this section,we consider two indicators of remoteness, namely, distance to an all-weatherroad as well as distance to the commune People’s Committee. The distance toan all-weather road is an indicator of how well connected the household is toits immediate surroundings. A long distance to an all-weather road increasestransportation time and can make transportation of people, as well as ofagricultural products and other goods, very difficult during floods. The dis-tance to the commune People’s Committee is a meaningful proxy for remote-ness since the People’s Committee, the administrative centre of thecommune, tends to be better connected to the rest of Viet Nam than moreremote parts of the commune. Figure 13.9 shows the additional distance thatethnic minorities have to an all-weather road and to a People’s Committee.

The finding that ethnicminority households are, on average, more remotelylocated can simply be due to the fact that population density is lower in theseparts of the country. In order to rule this out, Figure 13.9 also presents theeffects of belonging to an ethnic minority, using Kinh households within

-0.50

0.51

1.52

2.53

2008 2010 2012 2014

Km

Year

(a) Distance to road

00.20.40.60.8

11.21.41.6

2008 2010 2012 2014

Km

Year

(b) Distance to people's committee

Non-Kinh effectWithin-province non-Kinh effect

Non-Kinh effectWithin-province non-Kinh effect

Figure 13.9 Additional distances for non-Kinh households by year, 2008–14Notes: The effect shown is the parameter estimate of an ethnic minority dummy regressed ondistance to road using year-specific regressions. The within-province fixed effects include a full setof province dummies. The lines represent 95% confidence intervals.Source: Authors’ calculations based on the VARHS database.

5 The first phase of P135 was implemented over 2001–5 and the second phase over 2006–10.Cuong, Tung, andWestbrook (2014) assess the second phase and find that minority households intargeted communes experienced a larger decline in poverty than those in control communes.

Ethnic Disadvantage

269

the same provinces for comparison. We do this by running a year-specificregression that includes province fixed effects, similarly to the regression inSection 13.4.1.The additional distance to an all-weather road is greater for minority house-

holds in all years. In 2008, the average additional distance was just above 2 kmfor the sample as a whole and just below 2 kmwhen controlling for provincialdifferences. This is a long distance: the average distance to an all-weather roadfor Kinh households was 3.1 km in 2008. However, the discrepancy has fallenover time. In 2014, the within-province effect was statistically indistinguish-able from ‘0’. So while non-Kinh households still, on average, live furtheraway from roads than the ethnic Vietnamese, the entirety of this effect can, inlater years, be attributed to non-Kinh households living in provinces where allhouseholds—Kinh and non-Kinh—tend to live in more remote locations.The additional distance to the People’s Committee is also positive for the

minorities in all years. In 2008, the total additional distance was around0.8 km, or 0.25 km using the within-province estimate. The average distancefor Kinh households in 2008 was 2.4 km. The additional distance to thecommune People’s Committee is therefore smaller in both absolute and rela-tive terms, compared to the additional distance to roads that non-Kinh house-holds experience. The trend over time is less clear, but we do note that, as wasthe case with the distance-to-roadmeasure, the estimate of additional distancefor the non-Kinh in 2014 is statistically indistinguishable from ‘0’, onceprovince fixed effects are taken into account.

13.4.4 Social Networks

The final aspect of constraints we investigate is the social network of ethnicminority households. Figure 13.10 explores the extent of social segregation ofKinh and non-Kinh households by exploiting information on the ethnicity ofthe three most important people that a household can contact for money incase of emergencies. This, combined with information on the commune-levelshare of minority households, allows us to compare the share of contacts ofKinh ethnicity in a household’s network with the share of Kinh ethnic house-holds in the commune. If ethnicity does not play a role in the formation ofcontacts, one would expect these two shares to be equal. Figure 13.10 showsthe difference between these two shares for Kinh and non-Kinh households.The positive number for Kinh households implies that they have more con-tacts among other Kinh households than would be expected if the share ofcontacts was to mirror the share of Kinh households in the commune. Like-wise, the negative number for non-Kinh households implies that they havefewer contacts among Kinh households (and, therefore, more contacts among


270

other minority households) than expected if ethnicity did not play a role incontact formation.

This is, therefore, evidence of segmentation among ethnic lines. There areno indications that this discrepancy reduces over time; if anything, it appearsthat minority households get further isolated towards the end of the studyperiod. This is potentially problematic for the ethnic minority households,given that these ties may be less valuable in times of emergency, since, asshown in Section 13.1, ethnic minorities tend to be worse off and the linksmay therefore be less valuable. Further research is needed in order to under-stand how these links are formed and what the consequence of this differenceis for welfare outcomes.

13.5 Differences within the Non-Kinh

We now explore differences within the non-Kinh minorities. While we haveso far considered the non-Kinh as a homogeneous group, the fact remains thatoutcomes within non-Kinh households vary on account of differences such as

–20

–15

–10

–5

0

5

10

2008 2010 2012 2014

YearYearKinh Non-Kinh

Figure 13.10 Over- and under-representation of links to Kinh farmers compared to thecommune average, 2008–14Notes: Households were asked to name up to three contacts that they depend on for money in caseof emergencies. The figure shows the average share of those links in percentage that householdshave to Kinh households, minus the average share of Kinh households in the commune. If a groupscores 10 it means that the group has 10 percentage points more links to Kinh farmers than whatwould be expected if link formation was random. This could happen if communes consist of 75%Kinh farmers and 85% of these farmers’ links are to other Kinh farmers. Since this figure also usesdata from the commune questionnaire, the sample is somewhat reduced (N=2162 on average peryear). The share of Kinh households in the commune is only available for 2014. This share isassumed to be constant over time and is used for all years.Source: Authors’ calculations based on the VARHS database.

Ethnic Disadvantage

271

region they reside in, the specific ethnic group they belong to, and whetherthey know Vietnamese. In this section, we examine these three dimensions.As discussed in Section 13.1, non-Kinh minorities are largely concentrated

in the Northern Mountains and the Central Highlands of Viet Nam. Whileboth these regions are mountainous and have relatively limited access topublic services and infrastructure, previous research has noted that the minor-ities in the Central Highlands performed worse as compared to the NorthernMountain minorities during the 1990s (Baulch et al. 2007).Using the VARHS data, we compare the economic welfare of minority

households residing in the mountainous Northern region (provinces of LaoCai, Phu Tho, Dien Bien, and Lai Chau) to those residing in the CentralHighlands (provinces of Dak Lak, Dak Nong, and Lam Dong). Figure 13.11(a) shows how real per capita monthly food consumption evolved for

0

50

100

150

200

250

300

350

2008 2010 2012 2014

Tho

usan

ds

of

real

VN

D

Year

(a) Region

North Central

0

50

100

150

200

250

300

350

400

2008 2010 2012 2014

Thus

and

s o

f re

al V

ND

Year

(b) Ethnicity

Mu'ong H'Mong Tay Thai

0

50

100

150

200

250

300

350

400

2008 2010 2012 2014

Tho

usan

ds

of

real

VN

D

Year

(c) Speak Vietnamese

Yes No

Figure 13.11 Differences in monthly per capita food expenditures in real 1000 VNDwithin minorities by (a) region, (b) ethnicity, and (c) language, 2008–14Note: In (a) the Northern region includes the provinces of Lao Cai, Phu Tho, Dien Bien, and LaiChau and the Central region includes the provinces of Dak Lak, Dak Nong, and Lam Dong.Source: Authors’ calculations based on the VARHS database.


272

minority households over the period 2008–14. We find that, while there wereno regional disparities in 2008, the per capita consumption of minorityhouseholds in the Central Highlands grew a lot faster than that of the non-Kinh in the NorthernMountains. This finding is consistent with those derivedfrom the Viet Nam Household Living Standards Survey (VHLSS) data that theCentral Highlands experienced a higher reduction in poverty rates in the2000s (World Bank 2012).

Next, we examine whether the growth trajectories of minority groups varyby ethnicity. The non-Kinh consist of fifty-three officially recognized ethnicgroups. In order to have meaningful results, we limit our analysis to thoseminority ethnic groups where we have at least forty-five observations in eachwave of the VARHS data. This gives us four groups: Tay (the largest minoritygroup in Viet Nam), Thai, Muong, and H’Mong. As all four of these ethnicgroups largely reside in the Northern Mountains, comparative data can alsoshed further light on the economic stagnation among minority householdsdiscussed in the preceding paragraph.

In Figure 13.11 (b), we examine how real per capita monthly food consump-tion has evolved over 2008–11 for these four groups. We find that the Muongare consistently strong performers and the H’Mong consistently lag behindthroughout this period. On the other hand, the Thai and Tay exhibit a lot ofdynamics during this time period. While consumption rates of the Tay weresimilar to that of the Thai in 2008, and appear to have stagnated in 2010,consumption grew rapidly since then and was significantly higher than thatof the Thai in 2014 (p-value=0.025).

Finally, we consider knowledge and fluency in Vietnamese. Many of theminority groups either do not know or are not fluent in the Vietnameselanguage. In the VARHS data, 72.5 per cent of the 429 non-Kinh householdsinterviewed in 2014 reported that Vietnamese was not their main language.A lack of knowledge of the Vietnamese languagemay be preventingminoritiesfrom applying for credit, taking part in market transactions, and migrating,and may prompt them to drop out of school. This may also limit theirunderstanding of government programmes available in the commune thatare mostly in Vietnamese, leading to lower participation in such schemes.Indeed, as shown in Figure 13.11 (c), we find that minority households thatspeak Vietnamese as their main language are significantly better off than thosethat do not. Given the consistent nature of this finding, it is imperative toexplore the ways in which the lack of fluency in Vietnamese is restraining thegrowth of non-Kinh households and compounding the disadvantage theyalready face. A recent step in this direction has been a pilot programme toprovide mother tongue-based bilingual education to ethnic minorities in aneffort to improve learning outcomes for the non-Kinh and reduce school drop-out rates (UNICEF 2011).

Ethnic Disadvantage

273

13.6 Conclusion

Over the years, the government of Viet Nam has undertaken various measuresto address the ethnicity gap in Viet Nam. This includes setting up theCommittee for Ethnic Minority and Mountainous Area Affairs (CEMA) andspecifically targeting poverty in remote and inaccessible areas under policiessuch as the Socio-Economic Development Program for the Communes FacingGreatest Hardships in the Ethnic Minority and Mountainous Areas (P135).Over the period 2006–14, the average rural Vietnamese household in the

VARHS survey has seen spectacular improvements in living standards asmeasured by household income and consumption expenditure. However,national averages mask substantial differences in the level of welfare betweenthe Kinhmajority households and the non-Kinh households belonging to oneof Viet Nam’s fifty-three ethnic minority groups. Both groups have seenincreases in their living standards, but a significant difference in the relativelevel of welfare remains. In this period, there are no signs of convergence inwelfare between the two groups: the relative difference in food expenditureand household income in 2014 is almost identical to the difference observedin 2006. Similarly, on other indicators the evidence continues to be worri-some: while housing indicators of Kinh households have improved, they haveremained more or less stagnant for the average non-Kinh household.An examination of the sources of income reveals that the non-Kinh are

more likely to diversify into non-farm activities, but when non-Kinh house-holds do diversify, they are more likely to depend on CPR as opposed toKinh households that primarily engage in wage employment and householdenterprises. We also identified several constraints to help explain thesedifferences. Non-Kinh households have lower quality agricultural land andlower rates of ownership certificates. They also face more problems produ-cing and selling their agricultural products and have worse access to formaland informal credit. While remoteness was found to matter in the earlierpart of the period, non-Kinh households no longer appear to be moreremotely located than their Kinh counterparts living in the same provinces.There is, however, some evidence of segmentation in social networks alongethnic lines.Finally, we find a fair amount of heterogeneity within the non-Kinh minor-

ities along spatial, ethnic, and linguistic lines. Minority households residing inthe Central Highlands have grown faster than those in the northern moun-tains; the Tay and the Muong have fared better than the Thai and H’Mong;and minority households that speak Vietnamese have done better that thosethat do not. Overall, while differences between the Kinh and the non-Kinhcontinue to exist, it appears that, currently, social distance rather than geo-graphical distance plays a greater role in the slow growth of the non-Kinh.


274

Acknowledgements

We thank the participants at a workshop at the Central Institute for EconomicManagement in Hanoi in May 2015 for their valuable comments on an early versionof this chapter.

References

Baulch, B., H. T. Pham, and B. Reilly (2012). ‘Decomposing the Ethnic Gap in RuralVietnam, 1993–2004’. Oxford Development Studies, 40(1): 87–117.

Baulch, B., T. T. K. Chuyen, D. Haughton, and J. Haughton (2007). ‘Ethnic MinorityDevelopment in Vietnam’. Journal of Development Studies, 43(7): 1151–76.

Beck, U., S. Singhal, and F. Tarp (2016). Coffee Price Volatility and Intra-Household LabourSupply: Evidence from Viet Nam. WIDER Working paper 2016/16. Helsinki:UNU-WIDER.

Cuong, N. V., P. D. Tung, and D. Westbrook (2014). ‘Do the Poorest Ethnic MinoritiesBenefit from a Large-Scale Poverty Reduction Program? Evidence from Vietnam’.Quarterly Review of Economics and Finance, 56: 3–14.

The Economist (2015). ‘Ethnic Minorities in Vietnam: Out of Sight’, 4 April. Available at:<http://www.economist.com/news/asia/21647653-continuing-grinding-poverty-vietnams-minority-regions-liability-communist-party-out>. Accessed 1 September2016.

Glewwe, P., M. Gragnolati, and H. Zaman (2002). ‘Who Gained from Vietnam’s Boomin the 1990s?’ Economic Development and Cultural Change, 50(4): 773–92.

Khai, L. D., C. Kinghan, C. Newman, and T. Talbot (2013). Non-Farm Income, Diversifi-cation andWelfare: Evidence from Rural Vietnam. CIEMWorking Paper. Hanoi: CentralInstitute for Economic Management (CIEM).

Oostendorp, R. H., T. Q. Trung, and N. T. Tung (2009). ‘The Changing Role of Non-Farm Household Enterprises in Vietnam’. World Development, 37(3), 632–44.

Pham, H. T. and B. Reilly (2009). ‘Ethnic Wage Inequality in Vietnam’. InternationalJournal of Manpower, 30(3), 192–219.

Singhal, S. and U. Beck (2015). Ethnic Disadvantage in Vietnam: Evidence Using PanelData. WIDER Working Paper 97/2015. Helsinki: UNU-WIDER.

UNICEF (United Nations Children's Fund) (2011). ‘Action Research onMother Tongue-Based Bilingual Education: Achieving Quality, Equitable Education’. Technicalreport, UNICEF.

Van de Walle, D. and D. Gunewardena (2001). ‘Sources of Ethnic Inequality in VietNam’. Journal of Development Economics, 65(1): 177–207.

World Bank (2009). Country Social Analysis: Ethnicity and Development in Vietnam.Summary Report. Washington DC: World Bank.

World Bank (2012). Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on PovertyReduction and the Emerging Challenges. 2012 Vietnam Poverty Assessment Report.Washington DC: World Bank.

Ethnic Disadvantage

275

http://www.economist.com/news/asia/21647653-continuing-grinding-poverty-vietnams-minority-regions-liability-communist-party-out

http://www.economist.com/news/asia/21647653-continuing-grinding-poverty-vietnams-minority-regions-liability-communist-party-out

Part IVLessons and Policy

14

Lessons Learnt and Policy Implications

Finn Tarp

14.1 Introduction

This volume started in Chapter 1 by setting the scene for one of the mostimpressive national performances in modern-day socioeconomic develop-ment. And Viet Nam has indeed made great strides in its effort to transitionfrom a centrally planned economy to a more market-based institutional andeconomic system. Chapter 2 provided, in turn, further background informa-tion on the quantitative dataset on which the detailed studies in Chapters3–13 are all based—the Viet Nam Access to Resources Household Survey(VARHS). To reiterate, the VARHS has, over the years, produced a unique5-wave panel data set, covering a balanced sample of 2,162 households from12 provinces, surveyed every 2 years from 2006 to 2014.1

Prominently in a broader perspective, the design, implementation, and useof the VARHS provide a highly relevant case study of what the global call for adata revolution—subscribed to by the United Nations—means in actual prac-tice in the context of the 2030 sustainable development agenda. This is soespecially when the international discourse is tuned in on the ambition of‘leaving no one behind’. Arguably, Viet Nam is a particularly instructivecountry to study and learn from. Not long ago Viet Nam had an economicstructure that was comparable to that of many African countries today; andmuch has been realized in this still relatively poor, yet dynamic SoutheastAsian country since Doi Moi began in 1986, just thirty years ago.

Viet Nam admittedly sets a high bar when it comes to socioeconomicachievements. As such, it illustrates aspirations other countries may wish toadopt and even target. The Viet Nam experience certainly carries with it a

1 The data set is freely available from the following webpage: https://www.wider.unu.edu/VARHS.

series of lessons about how the development process can be managed. Import-antly, while impressive, we have argued throughout in this volume thatViet Nam could do even better moving forward. This is not because thegroup of authors who have contributed to this book subscribe to the, para-doxically, rather pessimistic outlook and world view one often encounters insocial and professional interactions in Viet Nam. Instead, it is derived fromcareful analysis carried out to uncover the elements of what remains to bedone, or, put better, what could be done in the next step of a never fullypredictable development process. Viet Nam is indeed a rising dragon on themove. Nonetheless, many challenges continue to lie ahead. A stated aim ofthis volume was to help identify these challenges more specifically, to get toknow them in-depth, and to reflect on what the key policy implications are forViet Nam and beyond, to other developing countries. Summarizing and out-lining these implications is the purpose of this concluding chapter.Chapter 1 provided institutional background and reviewed the macroeco-

nomic situation and development progress of Viet Nam in a comparativeregional perspective as an antecedent to the core chapters of the book. Basedon standard data available from international sources, I noted the stellarperformance Viet Nam has shown when it comes to the reduction of poverty,based on a robust process of transformative economic growth. The experienceof Viet Nam strongly confirms the key role growth and transformation hasto play in meaningful development. I also noted the active macroeconomicpolicy stance the Vietnamese government has taken in the face of the globalfinancial economic crisis, which has hit much harder in other developingcountries. Viet Nam has—as demonstrated in this volume—done well to takean active stance and respond. Other countries are well advised to take note ofthe fact that passive adaptation to exogenous events and influences is not theway to go. Surely an eye needs to be kept on avoiding ‘overdoing’ it andgradually getting trapped into vicious circles of public debt and similar issues.At this point it is my assessment that such problems are not on the horizonin Viet Nam. I, furthermore, believe these broad lessons should be carefullyconsidered and adapted as required to specific national circumstances across thedeveloping world more generally, and in the developed world as well!At the same time, themacroeconomic overview in Chapter 1 identified early

on a few thorny characteristics which seem to reflect underlying sector andmicroeconomic issues to which I return in Section 14.2: (i) agriculture valueadded per worker appeared, based on the international data, available to haveremained stagnant in Viet Nam over the past decade; (ii) information tech-nology (IT) development (as measured by fixed-wired broadband subscrip-tions) is far from impressive in regional comparisons; and (iii) while theprevalence of undernourishment and the depth of the food deficit are indeeddropping, progress elsewhere suggests more can be done.

Lessons and Policy

280

Following on from Chapters 1 and 2 we proceeded to the core of thevolume, and in eleven chapters we addressed three broad themes in socio-economic development:

• the process of rural institutional and economic transformation as itimpacts on almost all aspects of rural life and economic activities;

• the critical importance of household access to markets for land, labour,and capital (i.e. key production factors) and the importance of associatedinstitutions; and

• the ultimate welfare outcomes and distributional issues (among, forexample, households, genders, and ethnic groups). In the final analysisthis is the lens through which ordinary people, policy makers, andresearchers will all have to look to evaluate whether development policyand strategy are succeeding or not.

I turn to these three themes, or parts, one by one in Sections 14.2, 14.3, and14.4, before my final remarks and observations in Section 14.5. To put thisinto perspective, I note upfront that these three themes are closely related tothree core development challenges: structural transformation, inclusion andsustainability, and linked to the cross-cutting issues of development financeand gender. These topics are central both to the 2030 sustainable develop-ment agenda, the mandate of United Nations University World Institute forDevelopment Economics Research (UNU-WIDER), and indeed the futuredevelopment of Viet Nam; so the Viet Nam experience carries, as noted, aseries of implications which policy makers in other countries and in theinternational development community should consider carefully.

14.2 Rural Transformation

Development is interlinked with structural and institutional transformationof the economy, in the balance between its sectors, and in the nature ofeconomic activity of its people and the gradually evolving institutionswhich influence their behaviour. If the economy does not transform andinstitutions do not adapt progress is stifled and economic collapse mayensue. This volume included three studies of these processes. In Part I wefirst asked in Chapter 3 which insights the VARHS commune questionnairefurnishes as a complement to the macroeconomic picture discussed inChapter 1. We then proceeded in Chapter 4 to investigate the nature andcharacteristics of the ongoing diversification, commercialization, and trans-formation of the agriculture sector relying on household-level and institu-tional information. Thirdly and finally we analysed in Chapter 5 what ishappening in the non-farm rural economy. Non-farm activity will, in the


281

years to come, have to absorb an increasing share of the labour force ifdevelopment is to proceed and succeed.Chapter 3 documented that there are clear signs of transformation at the

commune level over time, especially in the provision of public facilities andinfrastructure. Public action matters, and this insight is a clear lesson also inother contexts. At the same time, significant regional differences exist, withthe Northern region lagging behind on several critical indicators. This fact,which cannot be captured by aggregate indicators for the economy as awhole, is critically important, and highlights the key and active attentionthe pursuit of spatial balance must play in less fortunate circumstances thanthose of Viet Nam. To be sure, Viet Nam continues to face the challenge ofensuring that its development becomes more regionally balanced, and this achallenge that is at the root of social upheaval and unrest elsewhere in theworld. This topic must therefore be centrally placed in any meaningful 2030agenda at national levels.Looking ahead, it also emerges from VARHS that commune leaders expect

climate change to become an important problem in the years to come.Speeding up development and the flexibility of the economy as traditionallyconceived is often the best available adaption strategy. A better educatedpopulation will know how to take proper action and will be better placed todo so. In Viet Nam, and elsewhere, this means that the basic policy message is:keep up the momentum. Accelerated development is not only desirable, it is akey tool to help peoples and nations to adapt effectively. It is, at the sametime, necessary to put on the thinking cap. Additional policy measures arerequired to help prepare farmers and other people for the many changes thatwill occur. In some cases this means policies should be put in place such thatrural people—or rather their children and grandchildren—will eventually, inthe longer run, be prepared and able to move to less affected areas and earn aproductive living there, less dependent on agriculture and existing flood-prone urban areas. I believe the policy lessons for other countries are indeedquite similar. This implies, for example, that forward-looking investmentprogramming, education, and regional development policies must be muchmore explicitly designed with climate change in mind.For the time being, agriculture continues, as demonstrated in Chapter 4, to

play a critically important institutional and economic role in rural Viet Nam.This is why attention to value added per worker in the rural sector is a criticalindicator. While the macro-evidence, provided in Chapter 1, is gloomy in thisregard, the more specific data and sources of information referred to inChapter 4 painted a somewhat more positive picture and interpretation. Fur-ther analytical work on this topic is therefore evidently needed in Viet Nam asis the case elsewhere. Chapter 4 also highlighted that participation in non-agricultural activities has been increasing over time. This is promising and

Lessons and Policy

282

reflects that development is happening, and that agriculture is increasinglybecoming commercialized in rural Viet Nam, mainly, but certainly not only,in the case of selling rice. It is notable, though, that while poorer householdsgrow more rice than previously, they sell less (due to more self-consumption).This is a sign that more needs to be done through well-designed policy andinstitutional development to better integrate such households into themarketeconomy and its institutions. This is, in turn, also the way to avoid marginal-izing groups of people to which I return in the remainder of this section. Thelessons for other countries, such as those on the African continent, whichhave, in many cases, paid much less targeted attention to agricultural devel-opment, are clear and persuasive.2

Other characteristics uncovered in Chapter 4 are that cash crops are animportant source of income, especially for households in the Central High-lands (i.e. coffee) and for better-off households, and that participation inaquaculture fluctuates from year to year, mainly due to its uncertain potentialfor income generation. This reinforces the points made in this section. It alsounderpins the finding that the VARHS data show a strong association betweencommercialization and wealth. Admittedly, cause and effect goes both wayshere. Yet it is certainly the case that increased commercialization of agricul-tural activities in rural Viet Nam has been an important contributor to theimpressive rural poverty reduction the country has experienced. This processneeds to be extended and completed to include those areas not yet fullycovered. The general lesson is that development policy and strategy had betterfocus on the development of effective market institutions, alongside an expli-cit goal of furthering production for sale (domestically and internationally).3

Chapter 5 put focus on the non-farm rural economy, and the VARHS dataconfirm the macroeconomic story of structural transformation in Viet Nam.At the micro level one observes that rural households are indeed shiftinglabour from agriculture towards operating household enterprises andengaging in waged labour outside the home. This is assuring, especially so asdiversification is found to be welfare improving and because the data showthat moving into operating household enterprises is an effective way forward.But the success of entrepreneurial activities hinges on appropriate access tofinance, quality training and education, and effective market access andintegration.

An important policy implication that goes well beyond the specific contextof Viet Nam is therefore that the role for the government is to shape anenvironment that will help generate and cultivate enterprises both in terms

2 See Arndt, McKay, and Tarp (2016) for a comprehensive elaboration of this argument.3 This insight is also emerging clearly from the research on industrial development that has

emerged in recent years, see, for example, Newman et al. (2016).


283

of ease of starting a business and in promoting access to, for example, creditbased on solid financial and economic cost-benefit analysis. At the householdlevel this has to be supplemented with diversification into waged employ-ment, which is an important source of welfare gain. Accordingly, another keypolicy priority is to help enable job creation, particularly in rural areas, forthose leaving agricultural production. And, fundamentally, the quality andquantity of education available and the quality and reach of existing infra-structure will, in the final analysis, be critical for the ability of families to findsuitable jobs outside agriculture.

14.3 Access to Resources

Four dimensions of households’ access to resources and associated institu-tional developments were covered in Part II: land and land markets, labourand migration, technology and innovation, and social capital and politicalconnections. While not exhaustive, these are key issues to address in comingto grips with production efficiency in the agricultural sector and how house-holds respond to their socioeconomic and institutional environment. It ishoped that the lessons learnt from these chapters will find their way intopolicy development elsewhere in the developing world.Chapter 6 showed that landlessness is not increasing and is positively cor-

related with income. Moreover, while farms are getting slightly smaller, plotsare being consolidated. The mean number of plots operated dropped from 5.8in 2006 to 4.1 in 2014, and there was a moderate increase in median plot size.This suggests that intra-farm consolidation rather than inter-farm consolida-tion is going on. This observation implies that more needs to be done topromote the critical role of market-based transactions for land; an insightthat has a wide-ranging set of implications.Land sales market activity is stagnant, not increasing, and there are reveal-

ing regional imbalances here: sales markets are much more active in theCentral Highlands than elsewhere. This suggests that a drive to promoteland markets elsewhere is called for. The potential is there, since the datashow that the activity in land rental markets is increasing. Rental marketstransfer land from rich to poor, and this is likely to increase efficiency, whichis encouraging. It is, however, important to ensure that farmers will be in aposition to own/control the land they are tilling. It is, according to thisperspective, worrisome that a significant number of plots still remain withouta red book (land-use certificate or LUC). The data reveal again regional vari-ation, with titling being least developed in the Northern Uplands. LUCs havea positive effect on private investment such as irrigation, and this effect issignificant and strong in the highland regions—where titling is least common.

Lessons and Policy

284

The policy implication is that titling should be expanded in the NorthernUpland and Central Highland regions, keeping a close eye onmonitoring thatthis does not lead to landlessness among the poorest. The lack of progress inthis policy area is, by now, likely to be a constraint to growth in value addedper worker in the agriculture sector, and it is an illustrative example of theneed for coming to grips with how institutions need to adapt to underpinfurther economic progress; a topic of widespread importance in many coun-tries aiming to move out of low-income status.

Turning to issues of labour and migration, Chapter 7 demonstrates signifi-cant movements of household members, both intra-province and inter-province. Some 20 per cent of the 2,162 interviewed households have at least1 member who migrated. The main reasons for migrating are, as expected,education and work-related motives. The VARHS also shows that, in the faceof shocks that threaten household welfare, remittances act as an importantshock-coping mechanism and channel for poverty reduction. It emerges aswell that better-off households are more likely to migrate. Importantly, thisindicates that there are constraints to migration for poorer households.

These insights mean that a key policy implication is that existing con-straints on voluntary migration should be removed, or at least made moreflexible, particularly for poorer households; a finding that is of widespreadrelevance elsewhere as well. Members in such households may have the desireto leave their home community to find productive work elsewhere, but maynot have the resources and possibilities to do so. Finally, while more specula-tive at this stage, there may well be a role for government or other agencies indeveloping formal banking mechanisms to facilitate the remittance of fundsback to sending households.

Land and labour are important production factors in the agriculture sector.So is information and communication technology (ICT), as discussed inChapter 8. While focused on ICT, this chapter hints in the conclusion thatmechanization of agriculture should remain a policy concern; an observationthat is in line with the need referred to elsewhere to boost agricultural product-ivity. In parallel, while we note: (i) a rapid increase in the ownership of mobilephones (households have moved from a median of zero to two phones in justeight years); and (ii) a large increase in internet access and computer ownership,this progress is not particularly significant in international comparison.

Moreover, the VARHS provinces lag behind the national average, and in2014, households that did not own phones were much more likely to befemale-headed, poor, and less educated than households with phones. Thisis a telling caution that policies should harness the potential of IT services andthe Internet to provide information and education, especially in remote andpoorer regions and to disadvantaged people and families. Another associatedpolicy that merits attention is the use of IT in promoting e-governance, which


285

is a topic that is likely to become of increasing importance in Viet Nam in thecoming years.It is not only physical and human capital that are important to growth and

productivity. The same goes for social capital, as validated in Chapter 9, wherea variety of social, political, and institutional characteristics and issues werebrought to the fore. As expected in a rural setting, family ties play an import-ant role in economic transactions and provide safety nets (informal insur-ance). Such links are worth furthering and supporting to the extent feasible ina dynamic process of change that will often be unsettling. Furthermore,Chapter 9 shows that household income is strongly and positively associatedwith being a member of the Communist Party, all else being equal. Thisconfirms that patronage relations would seem to be important in Vietnamesepolitics and social and economic interactions, in spite of strong public com-mitment to principles of equity.These findings highlight the critical importance of social and economic

policies geared towards furthering fair, predictable, and transparent socioeco-nomic principles where all members of society are subject to the rule of law,and where those in positions of political and economic power are heldaccountable for their actions. There is an evident parallel here in ongoingdebates in Viet Nam about the widespread problem of corruption, whichforms part of this set of issues. A cancer can threaten the structure and well-being of a human body, and even its life if proper care and treatment is notinitiated in time. In socioeconomic contexts, similar dangers exist if corrup-tion and societal values are allowed to degrade, leading to institutional decayand a vicious circle that undermines development.It is against this background that these highly relevant topics and their

policy implications have, in recent years, attracted increasing attention indevelopment policy analysis. Moreover, an important finding from theVARHS is that the subjective wellbeing of Vietnamese rural people is farbelow the levels one might expect, given the general economic progress.This should give reason for pause and reflection, and it stresses both theneed for further research and the crucial importance of adopting integratedapproaches to development policy and strategy, reflecting particular nationalcharacteristics and possibilities. This topic is closely related as well to the topicof welfare outcomes, to which I now turn.

14.4 Welfare Outcomes

The ultimate indicator of whether a society (rich or poor), and the economicand institutional policies it pursues, is succeeding, is whether more and betterways of living and welfare for all its citizens are furthered and generated. It is

Lessons and Policy

286

widely appreciated that welfare economics is no easy and straightforward fieldof economics. Professor Amartya Sen was awarded the Nobel Prize ‘for hiscontributions to welfare economics’; he reminds us that we need to go beyondwhat he terms ‘welfarism’.4 Welfare theory must be based on more thanindividual utilities, whether they are interpreted as pleasure, as fulfilment, oras revealed preference. In other words, Professor Sen rightly emphasizes theneed to take a broader view. We have in VARHS followed this advice bycarefully addressing issues related to gender, children and youths, and ethni-city, in addition to the more standard topics of welfare dynamics and aggre-gate household inequality.

Focusing first on Chapter 10, we reiterate that the VARHS panel data providea unique opportunity to study the economic welfare of individual householdsover time. Overall, the data show that household welfare, as measured by: (i)food consumption; (ii) household income; and (iii) household ownership ofassets, has improved. Similarly, the number of households classified as pooraccording to the Ministry of Labour, Invalids and Social Affairs (MOLISA) hasdeclined over 2006–14. There is, however, considerable volatility over timeeven for per capita food consumption. Similarly, there is significant spatialvariation.

Notable factors that influence improvements in household welfare by theindicators listed are education and the presence of migrants in the household.On the other hand, belonging to an ethnic minority is significantly associatedwith smaller increases in food consumption and income. Clearly, the gainsrealized inwelfare inVietNamhavenot been equally shared across the country.The key policy message emerging is that while much has been achieved in VietNam in terms of growth andpoverty reduction, important challenges remain toensure truly inclusive progress in the years to come. Accordingly, subsequentchapters focus on gender, children and youths, and ethnic minorities; issueswhich are highly relevant in other national contexts as well.

Chapter 11, on gender and the empowerment of women, brings out clearlythe fact that while the welfare of female-headed households has improvedover time, they continue to be worse off and more vulnerable to incomeshocks than male-headed households. The VARHS data do show an increasein female empowerment (as measured by proportion of time spent in wagedemployment, whether women are involved in land management decisionswithin the household, andwhether land is jointly titled in a female householdmember’s name) over 2008–14. Moreover, female empowerment is stronglycorrelated with household food consumption. This indicates that femaleempowerment is a strongpathway to improvingwelfare. Thesefindings suggest

4 See Atkinson (1998).


287

that efforts to promote gender equality, through laws such as the Law onGender Equality (2006) and the Land Law (2003), should be stepped up toaddress the remaining inequities referred to in this section; they are illustrativeof the kinds of policy implications that are relevant across a wide range ofdevelopment contexts, further elaborated on by Grown, Addison, and Tarp(2016).Turning to Chapter 12, on youth and employment, the analysis depicts a

society that has made great progress towards improving child welfare. Overthe span of 2008–14, the health of children and young people has improved,and school attendance has increased, in particular for children above the ageof 10. There has also been a decrease in child labour, which is most notable forthe most vulnerable age group. The policy challenge looking forward is toensure a wider spread of these gains. While both girls and boys have experi-enced improvements in health and schooling outcomes, the VARHS showsthat boys benefited more than girls. Similarly, while wellbeing has increasedover time for both minority and non-minority groups, substantial differencesremain, particularly in terms of educational outcomes. The implication is thattargeted approaches to address the needs of girls and marginalized groupsacross the full policy spectrum from infrastructure facilities, information cam-paigns to focused class interventions, and the allocation of qualified teachers,are called for in Viet Nam—and, by implication, elsewhere.These observations are interlinked with Chapter 13, on ethnicity. Over the

period 2006–14, the non-Kinh population continued to lag behind Kinhhouseholds in terms of income and food consumption. While the gap hasnot widened, it is still there and has remained relatively constant. Moreover,the structure of household income varies significantly between the Kinh andthe non-Kinh. It also seems that the non-Kinh are more likely to diversify outof agriculture, and the non-Kinh households that do diversify are more likelyto depend on common property resources (CPR). This is different from Kinhhouseholds, which rely primarily on more stable wage employment andhousehold enterprises. Income from CPR is muchmore susceptible to exogen-ous events, including environment-related factors, and this finding under-lines the continued vulnerability of non-Kinh households.Remoteness no longer appears to be an important factor constraining the

growth of non-Kinh households. Yet they continue to have worse access toformal and informal credit, and social remoteness may also be a factor. TheVARHS does provide some evidence of segregation along ethnic lines. It is,furthermore, clear from the richness of the VARHS data that there exists a fairamount of heterogeneity also within the non-Kinh minorities along spatial,ethnic, and linguistic lines. Minority households residing in the CentralHighlands progressed faster than those in the Northern Mountains, the Tayand the Muong fared better than the Thai and Hmong, and minority

Lessons and Policy

288

households that speak Vietnamese did better than those that do not. Impli-cations for public action are clear and compelling.

Viet Nam has, over the years, undertaken a range of measures to address theethnicity gap. This includes setting up the Committee for EthnicMinority andMountainous Area Affairs (CEMA) and specifically targeting poverty in remoteand inaccessible areas under policies such as the Socioeconomic DevelopmentProgram for the Communes Facing Greatest Hardships in the Ethnic MinorityandMountainous Areas (Program 135 or P135). While the policies inherent inP135 are showing some effects, it is clear from the VARHS analysis that theyneed to be pursued vigorously and deepened in the years to come if ethnicdifferences are to disappear in future developments. Such policy needs aregenerally an illustration of what needs to be done elsewhere as well.

14.5 Final Remarks

In this chapter I have identified a series of the more important findings, andpolicy implications which emerge from the VARHS 2006–14 panel data andthe analyses contained in this volume. I recall that the aim of the VARHS wasto document the wellbeing of rural households in Viet Nam, focusing, inparticular, on access to and the use of productive resources. In concluding, itis important to recognize that a survey such as VARHS cannot provide a fullcoverage of all possible interpretations of the development process. There arebound to be different assessments emerging depending on whether absolute orrelative approaches and perspectives are relied on. For example, based onavailable data, most—if not all—would say that absolute poverty has certainlydeclined, and it would also appear that relative inequality has not worsenedsignificantly. This runs, however, counter to the widespread interpretation thatinequality and associated gaps are increasing very substantially as part of theadvances made in Viet Nam as a consequence of the respectable rate of aggre-gate growth realized, and which is commendable in international comparison.

It is important here to keep inmind that if an economy grows on average by6.9 per cent per annum then average income is doubled every ten years.Assuming all incomes increase by this average rate, this means that a personwho earned an income equivalent to US$1 a day in 1986 is today very close toearning US$8 a day. Almost three decades have passed since Doi Moi wasinitiated and average growth has indeed been quite close to 6.9 per cent peryear. In contrast, a personwho earned US$10 dollars a day in 1986 will by nowbe earning US$80. While relative inequality between these two people hasremained unchanged it would appear they have fared very differently. Andindeed they have in absolute terms, which is the difference that will ofteninfluence perceptions of what has happened. The gap between them has


289

widened very significantly. The ancient Greek philosopher Plato was alreadyaware of the relative nature of human insight, that is, that different interpret-ations of the same reality may exist—see his dialogue The Republic, writtenaround 380 BC.5

The VARHS departed from the premise that, while this is understood, it issensible to collect and analyse quantitative data on the real life and circum-stances of rural people. Indeed, this is what the international calls for a datarevolution imply. And what did we find?We found, first of all, that living conditions have, in general, improved for

the surveyed households in absolute terms. This is not consistently the caseacross all areas of the country, and across different population groups. Toillustrate, Lao Cai failed to make significant progress over the 2006–14 perioddue to the particular combination of disadvantageous characteristics identi-fied in Chapter 10. And this is so even if most other provinces, including someinitially poorer ones from the north-west, advanced significantly. The dataalso show that, even in provinces where average living conditions improved alot, the situation deteriorated for a substantial minority of households inalmost every case.Thus, while the aggregate VARHS story clearly confirms the interpretation

of Viet Nam as a country that has experienced very significant poverty reduc-tion in rural areas, this is not true (in absolute terms) for all. There aresignificant numbers of households for whom the situation has worsened. Itwas also shown in this volume that having a sufficient level of assets, includ-ing education, social capital, and productive assets, is associated with a greaterlikelihood of becoming better off, as does having more prime-age householdmembers (and fewer dependents). Similarly, facing shocks and being of non-Kinh ethnicity are significantly associated with large reductions in, forexample, food expenditure. The policy implication, which is clearly general-izable to a wide set of developing country contexts, is:

• Maintain a robust focus on the need for continued development ofphysical, human, and social capital, with particular attention to disad-vantaged provinces and ethnic minorities. If not, existing gaps will onlywiden.

An integral part of this approach is based on the observation that agriculturalvalue addedmay not be increasing in line with general economic advance. Thisis critical because major numbers of the Vietnamese population continue todepend on agriculture for their livelihoods. Moreover, a sign of developmentis that labour productivity should equalize across sectors. This implies:

5 A study of this topic in global context is Niño-Zarazúa, Roope, and Tarp (2016).

Lessons and Policy

290

• Deepen and extend policies to promote agricultural productivity. Theyare key and need to be further developed and extended to all regions ofthe country, and they include both more traditional measures such ashigh-yielding seeds, information, and extension, and mechanization aswell as the intensification of modern technology embedded in IT. Actionsto remove the constraints to the functioning of land markets are alsoneeded. The fact that there are so few land transactions, particularly incertain regions, is a challenge that needs to be addressed vigorously andthe same goes for the low levels of land titling in these regions.

Development and earning a higher income involves, as is clear from theVARHS, people moving out of agriculture. While the agriculture sector mustgrow in absolute terms it should, over time, fall in relative terms. Structuraltransformation is needed and much can be done to further this process, theimplication being:

• Support off-farm activities and the establishment of household enter-prises actively as an integral part of a balanced strategy to promote entre-preneurial activity across the whole economy, and do not shy away fromsupporting the development of a more flexible labour market, character-ized by increased mobility. This needs to be done keeping in mind theneed for job creation. Unnecessary constraints to enterprise growthshould be addressed with a view to promoting an efficient allocation ofresources, and, in rural areas, programmes to support promising start-upenterprises to grow and expand could be intensified.

The VARHS also reveals that much can still be done to improve genderbalance and invest in improved conditions for children and the youth. Trulyinclusive development implies:

• Follow-up on the commitment to promoting gender balance in all itsdimensions through guidance and effective support at all levels, includ-ing the development of profile role models in all aspects of socioeconomicand political aspects of the Vietnamese society. Without policy interven-tions targeted at the most vulnerable groups (women and girls, ethnicminorities, and disadvantaged regions) the existing gaps in welfare willonly get wider. This is particularly the case in relation to investment inhuman capital. Poorer education outcomes for these groups mean thatthey will be left even further behind in the years to come.

I maintain that these policy implications are also essential elements of aneffective adaptation strategy to future climate change. The implication is:

• Promote broad-based development and increased flexibility to adapt tochanging circumstances, involving both a decreased role of exposed rural


291

areas, increased employment in other sectors elsewhere, and careful ruraland urban planning to avoid locking the country into inoptimal invest-ment patterns which are vulnerable to the effects of climate changes.

Summing up, the VARHS has revealed that while Viet Nam is a rising dragonon the move, market-based institutions are yet to be fully developed, as is thecase, for example, in relation to land and land market transactions. Furtherprogress in this regard—and in the many other dimensions of institutionalprogress to which reference has been made throughout this volume and inthese concluding remarks—is critical. This will require a focus on fair, predict-able, and transparent socioeconomic principles and practices where all mem-bers of society are subject to the rule of law, and where those in positions ofpower and influence are held accountable for their actions. A strong focus ondeveloping access to the Internet and promoting e-governance may be onespecific avenue to help this happen in practice.To conclude this volume, it is pertinent to recall that Viet Nam has, over the

past thirty years, grown from a very low level since the crisis of the mid-1980s.This means that Viet Nam has benefited from ‘low-hanging fruits’. It is widelyunderstood that once growth gets underway it is more easily sustained in low-income contexts, and this would certainly seem to form part of the relativesuccess Viet Nam has experienced compared to other countries worldwide.This volume has made an effort to come to grips with a range of important

development issues for the twelve provinces included in the VARHS dataset—and for Viet Nam as a whole—and I hope this effort will help inform policiesthat will lead to higher welfare and standards of living for future generationsin Viet Nam. From a more methodological point of view, this book hashighlighted the importance of carefully collecting data on the same house-holds over time to better understand the transformations occurring at thehousehold level.It is regularly argued by academics that policymakers are not always rational

and that they do not respond to research-based evidence. This is neither truenor constructive. Policy makers do respond—though not always in expectedways. Policy makers in Viet Nam have their goals—as is the case everywhereelse in the world—and pursue them with the available evidence at hand. It iswise to keep in mind here that John Maynard Keynes once stated:

Practical men who believe themselves to be quite exempt from any intellectualinfluence, are usually the slaves of some defunct economist. Madmen in authority,who hear voices in the air, are distilling their frenzy from some academic scribblerof a few years back.6

6 See <https://www.goodreads.com/quotes/212215-practical-men-who-believe-themselves-to-be-quite-exempt-from>. Accessed 8 August 2016.

Lessons and Policy

292

https://www.goodreads.com/quotes/212215-practical-men-who-believe-themselves-to-be-quite-exempt-from

https://www.goodreads.com/quotes/212215-practical-men-who-believe-themselves-to-be-quite-exempt-from

The VARHS was, as noted in the Preface, set up to help generate context-specific knowledge and evidence as well as increase analytical capacity. True,even without such evidence, policy makers will make decisions. They have to,but without research-based evidence they may not take the right decision. Itwas never the task of VARHS to tell the Vietnamese government what to do;the idea was to help provide analytical inputs—based on the construction of aunique panel data base—that will help improve policy-making. Looking to thefuture, one prediction is relatively certain: policy-making is not going to beeasier, it is going to be muchmore complex and demanding. This is one of thekey underlying reasons for the 2030 sustainable development agenda call for adata revolution—and I note that it is not often that the international com-munity calls for revolutions!

A final policy recommendation in this study is that Viet Nam would be welladvised to keep in mind the fact that low-hanging fruits are becoming scarce.This is an observation that reinforces the need to pay close policy attention tothe relatively low degree of subjective wellbeing of Vietnamese rural people.I hope the rising dragon and its leaders will have the stamina and wisdom tobuild constructively on the impressive achievements of the past, using theevidence in hand. If so, we will learn that the dragon was actually a real Asiantiger in disguise.

References

Arndt, C., A. McKay, and F. Tarp (2016). Growth and Poverty in Sub-Saharan Africa.WIDER Studies in Development Economics. Oxford: Oxford University Press. Avail-able at: <https://www.wider.unu.edu/publication/growth-and-poverty-sub-saharan-africa>. Accessed 15 May 2016.

Atkinson, A. B. (1998). ‘The Contributions of Amartya Sen to Welfare Economics’.Available at: <http://www.nuff.ox.ac.uk/users/atkinson/sen1998.pdf>. Accessed 15May 2016.

Grown, C., T. Addison, and F. Tarp (2016). ‘Aid for Gender Equality and Development:Lessons and Challenges’. Journal of International Development, 28 (3): 311–19. Avail-able at: <http://onlinelibrary.wiley.com/doi/10.1002/jid.3211/abstract>. Accessed8 August 2016.

Newman, C., J. Page, J. Rand, A. Shimeles, M. Söderbom, and F. Tarp (2016). Made inAfrica: Learning to Compete in Industry. Washington DC: Brookings Institution Press.

Niño-Zarazúa, M., L. Roope, and F. Tarp (2016). ‘Global Inequality: Relatively Lower,Absolutely Higher’. Forthcoming in Review of Income and Wealth; available at: http://onlinelibrary.wiley.com/doi/10.1111/roiw.12240/full.


293

https://www.wider.unu.edu/publication/growth-and-poverty-sub-saharan-africa

https://www.wider.unu.edu/publication/growth-and-poverty-sub-saharan-africa

http://www.nuff.ox.ac.uk/users/atkinson/sen1998.pdf

http://onlinelibrary.wiley.com/doi/10.1002/jid.3211/abstract

Index

accommodation and food services sector 96age

of children in household 249, 251of enterprise managers 96of household head 33, 35, 40, 45–6of household members 105

Agricultural and Rural Development SectorProgramme Support (ARDSPS) 30

agriculture 5–6, 54, 55, 98children and youths engaged in 238, 244,

248, 249–50, 252, 253choice 53–7collectivization 119–20and income 95and information and communication

technology (ICT) 176–7investment 132–4local structural transformation at

commune-level 53–7productivity 291resource allocation 21smallholder participation 72value 290–1value added per worker 5–6welfare dynamics 209–11see also commercialization in agriculture;

constraints to agricultural and non-agricultural production under ethnicdisadvantage

annual crops 133Apple 164aquaculture 54–66, 70–3, 74–80, 82, 84–8, 98Asian financial crisis (1997) 3assets of households 41, 44, 72, 92, 290

differential endowments 71ethnic disadvantage 260–2gender inequality and female

empowerment 225–6in households with children and youths 242social and political capital 199variables 199welfare dynamics 205–9, 215, 216–18,

219, 220

Bangladesh 223banking sector 10Bank for Social Policies (VBSP) 184bequests (of land) 122bonding social capital 181, 188, 190–1,

198, 200bridging social capital 181, 184, 188, 191,

198, 200bus stops, distance to 60, 61

Cambodia 2, 3agriculture 6current account balance per GDP 13domestic credit 11education level 19female life expectancy at birth 16food deficit 18foreign direct investment (FDI) 14GDP per capita growth 4government borrowing/debt 10household final consumption

expenditure 16inflation 9international reserves as share of GDP 14internet access 6labour force participation 8male life expectancy at birth 17population aged 15–64 years 7poverty headcount ratios 20sectoral distribution of production 5trade balance per GDP 13trade (exports plus imports) as share of

GDP 12under-5 mortality rates 17undernourishment prevalence 18

cash crops 56, 68, 70–1, 72, 73, 74–80, 82–4,87–8, 283

see also coffee productionchild labour 238–9, 243–4, 262children and youths 237–54, 288, 2916–9-year-olds 242–3, 244–5, 24610–14-year-olds 242–3, 245, 246, 248,

251, 253

children and youths (cont.)15–18-year-olds 242–3, 245, 246, 248age of children 249, 251agricultural activities 244, 248assets of households 242average number of children 239characteristics of households with

children 239–43child labour 238–9, 243–4cohort analysis 242–8days worked in agriculture 238, 249–50,

252, 253days worked outside the home 238, 249–50,

252, 253durable goods, ownership of 250economic status, improvements in 238education attendance and

attainment 238–9, 243–5, 246, 249–51,252–4

ethnic minorities/Kinh households 238–9,240, 242, 246, 247, 248, 254

female empowerment 251–3fertility rates: geographical variation 239–40,

242food expenditure per capita 242gender differences 238–9, 245, 248, 251, 254health 238–9, 243–4, 253–4household characteristics 248income, sources of 242land-use certificates (LUCs) 250number of days spent working 245–6panel study 248–53parents’ business activities 238proportion of households with children 240rural/urban area disparities 238shocks 251sickness incidence 245structural transformation 238, 253Viet Nam Access to Resources Household

Survey (VARHS) 237–40, 242, 243, 244Viet Nam Living Standard Survey

(VLSS) 238–9waged labour 238, 239, 244, 249–50, 252, 253welfare indicators 249–53wellbeing 238–9, 253–4

China 2agriculture 6current account balance per GDP 13domestic credit 11education level 19female life expectancy at birth 16food deficit 18foreign direct investment (FDI) 14GDP per capita growth 4government borrowing/debt 10household final consumption

expenditure 15, 16

inflation 9international reserves as share of GDP 14internet access 6labour force participation 8male life expectancy at birth 17population aged 15–64 years 7poverty headcount ratios 20sectoral distribution of production 5trade balance per GDP 13trade (exports plus imports) as share of

GDP 12trade/GDP ratio 10under-5 mortality rates 17undernourishment prevalence 18

civil registration (ho khau system) 141climate change 63–4, 66, 67, 291–2clinics as part of infrastructure 58–9, 63coffee production 70, 74, 82, 120, 123,

136, 283collective action problems 193–4commercialization in agriculture 68–88, 120

agricultural variables, outcomes for 82–4aquaculture 70–1, 72, 73, 74–80, 82, 84–8cash crops 68, 70–1, 72, 73, 74–80, 82–4,

87–8climatic conditions 74, 87coffee production 70, 74, 82commercial agricultural activities,

households engaged in 79–80determinants: in-depth analysis 80–7fisheries 72food crops 71, 72food security 69–70, 71geographic factors 76–7, 82, 87land use rights 69livestock 72production and sales dynamics 79–80relevant literature 71–2risk 71–2summary characteristics 73–4Viet Nam Access to Resources Household

Survey (VARHS) 70, 72–4, 87see also rice production

Committee for Ethnic Minority andMountainous Area Affairs(CEMA) 274, 289

common pool resources (CPR) 257, 263,274, 288

Communist Party membership 2, 182–4,193–4, 196, 198, 200, 286

Community-Based Monitoring System(CBMS) 43–4

computer ownership 160–2, 167–71, 177construction sector 54, 56, 66, 97–8consumer price index (CPI) 7, 207, 209consumption per capita 92, 107cooperatives 69

Index

296

corruption 117, 128, 286credit, access to 72, 82, 92, 267–8

information and communication technology(ICT) 176–7

labour and migration 154–6credit markets 21credit rationing 21crops

annual 56, 133choice restrictions 132–3food 71–2perennial 56, 131–3, 136seeds and fertilizers 93see also cash crops; coffee production; rice

productioncurrent account balance per GDP 12–13

deforestation 55, 57, 129Demographic and Health Surveys (DHS) 43, 44Development Analysis Network 92diseases 63–5district center, distance to/access to 60, 61diversification 92, 93, 107, 237Doi Moi reforms 1–2, 26, 58, 68–9, 117,

119–20, 140, 256domestic credit 11

see also credit, access todrinking water distribution 60–2, 66drought 93durable goods, ownership of 250

e-commerce 168neconomic activities of individual household

members 98, 224, 228, 230–1economic development and social and political

capital 181economic mobility 92economic welfare 287education attendance and attainment 17, 19,

33, 41, 44, 92, 93, 199, 290all household members 36, 37, 40children and youths 238–9, 243–5, 246,

249–51, 252–4enterprise managers/owners 93, 96ethnic disadvantage 257, 262gender inequality and female

empowerment 224, 228, 229–30, 235household heads 33, 36–9, 45–7information and communication technology

(ICT), access to 159, 166, 170, 174, 178level 17, 19migration 141, 156return to 57social and political capital 184, 199welfare dynamics 218–19, 220

e-governance 292emergency loans 191

engineering sector 97–8entrepreneurship 184environmental problems 64–5equity 91Ethiopia 92, 93ethnic disadvantage 105, 107, 256–74asset ownership rates by asset and

ethnicity 260–2child and adolescent labour 262children and youths 238–9, 240, 242, 246,

247, 248, 254constraints to agricultural and non-

agricultural production 264–71access to credit and borrowing 267–8additional distances travelled for non-Kinh

households 269–70land and agriculture 264–7remoteness 268–70social networks 270–1social segregation of Kinh and non-Kinh

households 270–1descriptive statistics by ethnicity 257–8differences within the non-Kinh 271–3Doi Moi reforms 256educational attainment/school

attendance 262food expenditure 257–9, 272–3, 274H’Mong 273, 274, 288Hoa 257household enterprises 262–3, 274household income 256, 257–60, 262–4, 274important constraints after harvest 266–7important constraints before harvest 266Khmer 256Kinh 256land quality and ‘red-book’

ownership 265–6literacy of household head 257Muong 256, 273, 274, 288non-farm activities 274non-Kinh 256social networks/social distance 274Tay 256, 273, 274, 288Thai 256, 273, 274, 288Viet Nam Access to Resources Household

Survey (VARHS) 257, 272–4Viet Nam Household Living Standards

Survey (VHLSS) 273wage employment 262–3, 274welfare dynamics 217–19, 258–62

ethnicity 33, 41, 199, 288–9see also ethnic disadvantage; Kinh ethnicity

EVN Telecom 159exclusivism 181expenditure of households 15, 16, 92extension centres/shops, distance to/access

to 60, 61, 63

Index

297

external economic relations 10–14current account balance per GDP 12–13domestic credit 11foreign direct investment (FDI) 12, 14international reserves as share of GDP 12, 14trade balance per GDP 12–13trade (exports plus imports) as share of

GDP 12US dollar/Vietnamese Dong exchange

rate 12, 15

facilities in communes, types of 58–9family relations/inheritance 93family ties 189–93, 200Farmers’ Union 184farm size 120–2female empowerment see gender inequality

and female empowermentfemale-headed households 170, 178, 199, 219,

223, 224–7female life expectancy at birth 15, 16fertility 41, 239–40, 242financial helpers 191–2, 194, 198fisheries 72see also aquaculture

Food and Agricultural Organization (FAO) 68Fisheries and Aquaculture Statistics

Database 54food consumption 206–9, 212–19, 233–4food crisis (2007/8) 68food crops 71–2food deficit 17, 18food expenditure 290children and youths, households with 242ethnic disadvantage 257–9, 272–3, 274migration 144–5, 151–4, 156rural non-farm economy 100–1

food security 69–70, 71foreign direct investment (FDI) 12, 14forestry 54, 55, 56, 98see also deforestation

gender 92, 93balance 291children andyouths 238–9, 245, 248, 251, 254of household heads 33, 34, 45–6, 47, 166see also gender inequality and female

empowermentGender Equality Law (2006) 222, 288gender inequality and female

empowerment 222–36, 251–3, 287–8assets 225–6cohort analysis 227–31

economic activities of individualhousehold members 224, 228, 230–1

education outcomes 224, 228, 229–30, 235health outcomes 224, 228–9

female-headed households 223, 224–7food consumption 233–4household income 224–7indicators 232–3, 235joint titling of land 222–3, 224, 235land-use certificate (LUC, ‘red book’) 226,

232–3Viet Nam Access to Resources Household

Survey (VARHS) 223, 224, 228, 232, 235Viet Nam Household Living Standards

Survey (VHLSS) 227–8vulnerability of female-headed

households 227welfare outcomes 231–4

General Statistics Office (GSO) 21, 27, 164,169, 207

geographical factors 66and commercialization in agriculture 76–7,

82, 87information and communication technology

(ICT) 160–2Ghana 93global financial crisis (2007–8) 3, 10government borrowing/debt 9–10government land expropriation 117government officials, connections

with 198, 200gross domestic product (GDP) 2–4

handicrafts sector 54, 55health care sector/health-care centres, access

to 44, 57–9, 63children and youths 238–9, 243–4, 253–4gender inequality and female

empowerment 224, 228–9hospitals as part of infrastructure 58–9household characteristics and labour/

migration 144–7household composition 218household consumption and socioeconomic

indicators 15–19education level 17, 19female life expectancy at birth 15, 16food deficit 17, 18household consumption expenditure 15, 16male life expectancy at birth 15, 17under-5 mortality rates 16, 17undernourishment prevalence 17, 18

household enterprises 93, 95, 98, 103, 107,288, 291

ethnic disadvantage 262–3, 274see also rural non-farm economy

household income 44, 72, 93, 102, 196–8, 200,289, 291

children and youths, householdswith 237, 242

diversification 101

Index

298

ethnic disadvantage 256, 257–60, 262–4,274, 288

gender inequality and femaleempowerment 224–7

increase 193–5and information and communication

technology (ICT) 159, 166, 170–1,174, 178

and landlessness 119levels and land use 57local structural transformation at commune-

level 52per capita 52–3rural non-farm economy 94–5, 100–1shocks 105, 107social and political capital 181, 184, 186,

188–9, 193–8, 200welfare dynamics 206–12, 214–19

household infrastructure 166household level 19–22household size 217housing situation 41

income see household incomeIndia 92, 223Indonesia 2, 3

agriculture 6current account balance per GDP 13domestic credit 11education level 19female life expectancy at birth 15, 16food deficit 18foreign direct investment (FDI) 14GDP per capita growth 4government borrowing/debt 10household consumption expenditure 16inflation 9international reserves as share of GDP 14internet access 6labour force participation 8male life expectancy at birth 17population aged 15–64 years 7poverty headcount ratios 19, 20sectoral distribution of production 5trade balance per GDP 13trade (exports plus imports) as share of


inflation 7–9informal networks 198, 200informal sector 95, 97Informal Sector Survey (ISS) 43informal ties 182information and communication technology

(ICT) 158–78, 285–6and agriculture 176–7

asymmetric digital subscriber line (ADSL)connections 169

computer ownership 160–2, 167–71, 177credit and insurance 176–7descriptive statistics 162–71

computer ownership and internetaccess 167–71

phone ownership 163–7determinants of adoption 171–6

internet access 174–6phone ownership 172–4

education 159, 166, 170, 174, 178female-headed households 170, 178fixed-line phone 160–1, 164gender of household head 166geographical differences 160–2government policy 177household income 159, 166, 170–1, 174, 178household infrastructure 166information gathering 168information sources 176–7infrastructure 159, 162, 171internet access 158–62, 168, 170, 172, 174–8market information 176–7market policy 176–7mobile phones 158–9, 160–1, 164no internet access 170no phone ownership 165other technology, ownership of 171,

174, 178phone ownership 163–7, 172–4, 177–8remoteness 166, 171, 174size of household 174social capital 167social networks 177structural transformation 158technological progress 158television ownership 160–1, 177trust measures 167, 171wealth 159

information costs 72information gathering 168infrastructure 72information and communication technology

(ICT) 159, 162, 171investments 63provision 57–63, 66

Institute of Labour Science and Social Affairs(ILSSA) 27

insurance, access to 72, 176–7, 193International Labour Organization (ILO) 7International Monetary Fund (IMF) 2international reserves 12, 14International Rice Research Institute (IRRI)

68, 75internet access 6, 62–3, 66, 158–62, 168, 170,

172, 174–8, 292

Index

299

irrigation systems 64–5and government investment 133–4, 136

Italy 187

joint titling of land 222–3, 224, 235

Kinh ethnicity 35, 36, 40, 45–6, 82, 145–6, 149

labour force participation 7–8Labour Force Survey (LFS) 43labour and income 95labour-leisure choice 71labour market 43labour and migration 139–56, 284, 285characteristics of migrants 147–8credit, access to 154–6data 141–4distribution of households by migration

status and food expenditure 144–5education 156food expenditure 144–5, 151–4, 156household characteristics 144–7

by migration status 145by reason of migration 146

international migration 142inter-province migration 142, 148, 156intra-province migration 142, 156Kinh ethnicity 145–6, 149networks, role of 148occupations of migrants 147–8permanent migration 142, 143–4policy background and literature

review 140–1province of destination 143province of origin 142–3reasons for migration 143–4remittance behaviour 149–51, 153–4shocks 146, 149, 151–3, 154, 155, 156temporary migration 142, 143–4wealth of household 144–5welfare of sending households 151–4working migrants 142–3, 144–5, 146, 147,

151–3, 155work-related motives 156

labour productivity 93land 92, 117–36agricultural investment 132–4annual crops 133bequests (of land) 122coffee production 120, 136collectivization of agriculture 119–20commercialized agriculture 120and credit markets 21crop choice restrictions 132–3economic circumstances 136farm size 120–2forest clearing 129

fragmentation 22, 120–2inter-farm consolidation 122, 135intra-farm consolidation 135irrigation and government

investment 133–4, 136Land Laws (1993, 2003, 2013) 69, 128,

222, 288land markets 66, 123–8, 284migration 135number of plots operated by region 121–2ownership 44perennial crops 131–3, 136property rights 128–31, 132–4, 136purchases and sales of land 123–5quality 265–6rental markets (land) 124–8, 136, 191rental market where tenant is relative of

landlord 192–3residential 55–6, 57, 66rice production 131, 132rural non-farm economy 121size of land 92–3tenure security 21use rights 69use shares by year and region 55–6, 66see also joint titling of land; land titling;

land-use certificate (LUC, ‘red book’)landlessness 93, 118–20, 135land titling 69, 117, 128–30, 223, 284–5

and effect on agricultural investment 130–5land-use certificate (LUC, ‘red book’) 69,

128–30, 131–3, 135, 136, 284children and youths, households with 250ethnic disadvantage 265–6gender inequality and female

empowerment 226, 232–3land-use plans 132land use shares 55–6, 66Lao PDR 4, 20life expectancy

female (at birth) 15, 16male (at birth) 15, 17

linking social capital see political capitallivestock ownership 260–1loans, government-sponsored 184local structural transformation at

commune-level 51–67agricultural choice 53–7commune problems in past, present, and

future 63–6, 67communes by income and region 52drinking water distribution 60–2, 66facilities in communes, types of 58–9households, number of (in VARHS

communes) 53infrastructure provision 57–63, 66internet access points 62–3, 66

Index

300

land use shares 55–6, 66occupational choice 53–7, 66public goods provision 57–63, 66streetlights, prevalence of 60–2, 66transportation and other facilities, distances

to 59–61, 66

macroeconomic and monetaryperformance 2–10

agriculture 5–6banking sector 10external macroeconomic performance 10government borrowing/debt 9–10gross domestic product (GDP) growth 3–4inflation 7–9internet access 6labour force participation 7–8macroeconomic policy 9monetary policy interest rate 7, 9population aged 15–64 years 6–7sectoral distribution of aggregate output 5

main roads (distance to) 60, 61, 63Malaysia

domestic credit provided by financialsector 11

female life expectancy at birth 16GDP per capita growth 4male life expectancy at birth 17poverty headcount ratios 19, 20

market failures 71market participation 71, 72markets 59, 63marriage 41mass organizations (MOs) 184–6, 194, 198, 200migration 39, 53, 135, 185

welfare dynamics 219, 220see also labour and migration

MillenniumDevelopmentGoals (MDGs) 43, 222Ministry of Labour, Invalids and Social Affairs

(MOLISA) 80, 219, 287mistrust 188–9monetary policy interest rate 7, 9monetary sector 7mortality 16, 17, 41Mozambique 93

natural disasters 63–5networks, role of 148

see also social networksNigeria 92non-agricultural livelihoods 70, 291

see also household enterprises; rural non-farm economy

non-mass organization membership/voluntaryorganizations, other 186–8, 194

occupational choice 53–7, 66

paid employment see waged employmentPeople’s Committee, distance to 269–70perennial crops 56, 131–3, 136Philippines 4, 20, 29phone ownership 158–9, 160–1, 163–7, 172–4political capital 181–2, 184see also social and political capital

political connections 284population aged 15–64 years 6–7population censuses 41, 44–7population increase 53post offices, distance to/access to 58–9primary schools as infrastructure 58–9processing and manufacturing sector 96, 98productivity 22, 69, 70, 72, 93property rights 128–31, 132–4, 136Provincial Competitiveness Index (PCI) 44public goods and infrastructure

provision 57–63, 66, 71‘pull’ factors 93purchasing power 93‘push’ factors 92, 93, 94

remittance behaviour and migration 149–51,153–4

remoteness and ethnic disadvantage 268–70residence, prior place of 41resources, access to 284–6rice production 55–6, 57, 66, 131, 132characteristics of households selling

compared to non-sellers 80–1children and youths in household 239commercialization in agriculture 69, 72–3,

82–4, 87cultivation and sales 74–80percentage of households growing 73, 74, 75percentage of households selling 73, 76–7

roads, distances to/access to 63, 64–5, 269–70rural non-farm economy 91–112description of non-farm activities 94–8determinants of transition out of

agriculture 105–6diversification, impact of on household

welfare 102–4, 110–12diversification and transition from

agriculture 98–101economic activities of households 94–5, 98–9empirical analysis 101–7external employment 96–7food expenditure per capita 100–1household enterprise characteristics 95–6household income 94–5, 100–1industry sectors of enterprise operation 108industry sectors of external employment 109summary statistics 109welfare measures 100–1

rural transformation 281–4

Index

301

savings 93secondary schools as infrastructure 59, 63sectoral distribution of production (Southeast

Asia) 5seeds and fertilizers 93shocks 93, 94, 141, 219, 251labour and migration 146, 149, 151–3, 154,

155, 156size of household 92, 174Small and Medium-Scaled Enterprise (SME)

Survey 44social capital 92, 167, 284, 286, 290social evils 64–5social inclusion 91social networks 177, 270–1, 274social and political capital 181–200asset variables 199bonding social capital 181, 188, 190–1,

198, 200bridging social capital 181, 184, 188, 191,

198, 200collective action problems 193–4Communist Party membership 182–4,

193–4, 196, 198, 200education 184, 199endogenous variables 194ethnicity 199exogenous variables 194–6family ties 189–93, 200female-headed household 199financial helpers 191–2, 194, 198government officials, connections with

198, 200household income 181, 184, 186, 188–9,

193–8, 200informal networks 198, 200insurance 193land rental market where tenant is relative of

landlord 192–3mass organizations (MOs) 184–6, 194,

198, 200non-mass organizationmembership/voluntary

organizations, other 186–8, 194private returns to social capital 193–9trust 188–9, 195, 198

generalized 194, 200social remoteness 288social segregation of Kinh and non-Kinh

households 270–1Socio-Economic Development Program for the

Communes Facing Greatest Hardships inthe Ethnic Minority and MountainousAreas (Program 135) 269, 274, 289

streetlights, prevalence of 60–2, 66structural transformation 57, 59, 107, 291children and youths, households with

238, 253

information and communication technology(ICT) 158

see also local structural transformation atcommune-level

sub-Saharan Africa 72sustainable development agenda 2030 279,

281, 293

technological progress/innovation 158, 284television ownership 160–1, 177tenure security 21Thailand 2, 3

agriculture 6current account balance per GDP 13domestic credit 11education level 19female life expectancy at birth 16food deficit 18foreign direct investment (FDI) 14GDP per capita growth 4government borrowing/debt 10household consumption expenditure 15, 16inflation 9international reserves as share of GDP 14internet access 6labour force participation 8male life expectancy at birth 17population aged 15–64 years 7poverty headcount ratios 19, 20sectoral distribution of production 5trade balance per GDP 13trade (exports plus imports) as share of


tourism sector 56trade/GDP ratio 10, 12–13transaction costs 71, 72transportation and other facilities, distances

to 59–61, 66, 266transport costs 72trust 167, 171, 188–9, 195, 198

generalized 188–90, 194, 200

undernourishment prevalence 17, 18United Nations 279United Nations University World Institute for

Development Economics Research(UNU-WIDER) 281

US dollar/Vietnamese Dong exchange rate12, 15

Veterans’ Union 184Viet Nam Access to Resources Household

Survey (VARHS) 22–3, 26–41, 279,282–3, 285, 289–91, 293

age of household head 33, 35, 40

Index

302

aquaculture 75–6cash crops 75–6characteristics of balanced sample 31–9children and youths 237–40, 242, 243, 244climate change 65Commune questionnaire 22–3, 31, 51–2, 281communes 30comparison with Viet NamHousehold Living

Standards Survey (VHLSS) andPopulation Census 44–7

educationall household members 36, 37, 40household heads 36–9

empirical analysis 101ethnic disadvantage 257, 272–4gender of household heads 33, 34gender inequality and female

empowerment 223, 224, 228, 232, 235household and communes (balanced

sample) 31, 32household level 21–2information and communication technology

(ICT) 159, 160, 164, 168–9, 176, 177–8Kinh ethnicity in households 35, 36, 40labour and migration 139, 141–2, 156land, credit, and labour 43land issues 117, 118, 120, 131, 132, 134, 135local transformation 51–2, 53, 55location of (twelve) provinces 29main household questionnaire 22–3map of Viet Nam 28non-attrition 40non-farm activities of households 94public goods and infrastructure 57, 58, 63purpose 26–7rice production 75–6rural non-farm economy 107sample attrition 39–41sample design 27–31sample size and survey months 27social and political capital 182, 184, 188,

189–91, 195welfare dynamics 205–7, 211, 219welfare outcomes 287youth and employment 288

Viet Nam Enterprise Census (VEC) 44Viet Nam Enterprise Survey (VES) 44Viet Nam Household Living Standards Survey

(VHLSS) 26, 27, 29, 31, 42–3communes 42community survey 42comparison with Viet Nam Access to

Resources Household Survey (VARHS)and Population Census 44–7

consumption poverty rates 43core questionnaire/questions 42enumeration areas (EAs) 42

ethnic disadvantage 273gender inequality and female

empowerment 227–8health, education, employment, household

assets, infrastructure 43household level 20labour and migration 139, 141living standard indicators 43province and urban/rural status 42rotating design 42–3welfare dynamics 205, 207, 220

Viet Nam Living Standards Survey (VLSS) 27,42, 205, 238–9

Viet Nam Post and Telecommunications Group(VNPT) 159

waged employment 98, 103, 107, 141, 262–3,274, 288

children and youths 238–9, 244, 249–50,252–4

welfare dynamics 209–11water distribution networks 63–5wealth of households 102, 144–5, 159welfare dynamics 100–1, 151–4, 205–20agricultural income 209–11assets 205–9, 215, 216–18, 219, 220attrition 215–16descriptive analysis of household welfare

changes 212–15econometric analysis of welfare

change 216–19education 218–19, 220ethnic minority groups 217–19female-headed households 219food consumption 206–9, 212–13, 214, 215,

216–19household composition 218household income 206–12, 214, 215,

216–18, 219household size 217household welfare measurement 206–9kernel density plots of welfare

measures 208–9migration 219, 220shocks, negative 219Viet Nam Access to Resources Household

Survey (VARHS) 205–7, 211, 219Viet Nam Household Living Standards

Survey (VHLSS) 205, 207, 220welfare measures, changes in 216–18wellbeing 219

welfare gains 107welfare indicators 249–53welfare levels and trends 258–62welfare outcomes 231–4, 286–9wellbeing 219, 238–9, 253–4wholesale and retail trade sector 96

Index

303

Women’s Union 181, 184work experience, prior 93World Bank 159, 244poverty line 19stabilization and structural programmes 2Viet Nam 2035 223, 238Viet Nam Country Gender Assessment 222World Development Indicators (WTI) 2

World Development Indicators (WDI)data 5

World Trade Organization (WTO)membership 2

World Values Survey (WVS) 43, 44, 189

Young Lives Project 43, 44Youth Union 184

Index

304